The human visual system
INTRODUCTION
Vision is perhaps the most remarkable of the senses. The human eye has been likened to a camera in many descriptions and indeed, superficially, this is true. As may be seen in Figure 4.1, it is a light-tight sphere with a lens system positioned with which to focus light on to a photosensitive layer, the retina, located at the back of the eyeball. Beyond this basic description, the structure and operation of the human visual system (HVS) is much more complicated than the most sophisticated cameras available at present. On examination, it may be seen that it is proficient in refined control of focusing, exposure, and white balance, capable of compression, is a scanning system and can respond to lighting conditions that can vary by up to six orders of magnitude. Furthermore, it can competently provide information on the three-dimensional world around us.
Figure 4.1 Cross-section through the human eyeball.
Adapted from McDonald (1987)
Generally, the HVS works so reliably on a daily basis that its complexity is forgotten, as are the myriad of processes that lead to our perception of the world about us. The sophisticated processing that takes place within the HVS, and predominantly within the visual cortex, leads us to perceive images approximately one- to two-tenths of a second after they occur. Understanding the basic functioning of the eye leads to better design and operation of imaging systems, whether it be matching the exit pupil of a pair of binoculars to that of the eye or designing a compression system to be perceptually lossless. The visual systems of animals display incredible variation in complexity, operation and performance. Appreciation of this wide biological diversity, from the compound eye of the bee to the polarized vision of cephalopods, inspires many further advances in numerous areas.
THE PHYSICAL STRUCTURE OF THE HUMAN EYE
In brief, the eye is a light-tight sphere, approximately 24 mm in diameter, whose shape is predominantly maintained by the sclera, the white part of the eye, and the vitreous humour (Figure 4.1). It has a lens system positioned at the front to focus light on to a photosensitive layer, the retina, which lines the rear of the eye, to form an inverted image. The lens system consists of the cornea and a crystalline lens. It is the function of the retina to convert the incoming light to electrical signals which then travel to the visual cortex and other structures via the optic nerve at the rear of the eyeball. Processing of images, their meaning and the context within which they appear is distributed throughout various parts of the brain and not presently fully understood. The visual cortex, however, is primarily responsible for perception of patterns and shapes encoded by the retina. The coloured portion of the eye, the iris, controls the amount of light entering the visual system by changing the size of the pupil, the dark centre, in a similar manner to a lens aperture. When combined with chemical changes in the retina this accounts for the eye’s ability to respond to varying lighting conditions. Considerable detail is omitted from the above overview and the remainder of this chapter is dedicated to describing the structure and processes of the human visual system at greater leisure.
Tunics
The eye may be considered as comprising three tunics, or layers, identified as the fibrous, vascular and nervous tunics. These approximately correspond to the outer, middle and inner layers of the wall of the eye. The fibrous tunic, or the outer layer of the eye, consists of the sclera and cornea. The sclera, also known as the sclerotic coat, is the ‘white’ of the eye, an opaque tough fibrous material made of collagen and elastic tissues. Protecting the rear three-quarters of the eyeball, it generally ‘yellows’ as it ages.
The middle, vascular tunic consists of the iris, ciliary body and choroid. Sometimes referred to as the uveal tunic, the primary function of this layer is to provide the eye with the oxygenated blood that it needs to operate. This is principally undertaken by the choroid, a network of blood vessels which lie in the posterior region of the eye below the sclera. The vascular tunic is pigmented with a deep purple colour, often referred to as ‘visual purple’, which absorbs scattered light in a similar manner to the matte-black coating often seen inside cameras. The inner, nervous tunic is the sensory layer containing the retina and upon which images are focused. The function of the layer is to detect light and encode the signal for transmission via the optic nerve to the visual cortex.
Cornea
Though its shape is fixed, the cornea provides approximately three-quarters of the optical power necessary to focus light on to the retina. It is a clear, avascular dome, approximately 12 mm in diameter and up to 1.2 mm thick, which covers the front part of the eye, the iris and the lens. A reasonable estimate of the refractive index of the cornea is of the order of 1.376. It is, however, constructed of many layers of fibrous tissue and, as such, its refractive index varies with each layer, hydration and the wavelength of light considered. Light passing through the cornea enters into the anterior chamber formed between its posterior surface and the iris. It is filled with aqueous humour which, necessarily, due to the avascular nature of the cornea, provides it with the nutrients it needs via diffusion.
The aqueous humour has a refractive index (see Chapter 2) of around 1.336 and, as such, the cornea–aqueous boundary provides no significant optical power. The majority of the optical power of the eye is therefore derived at the air–cornea interface. This is one of the reasons that when swimming underwater without a face mask or goggles human vision is blurred. The refractive index of water is approximately 1.335 at a wavelength of 550 nm as opposed to practically 1 for air. The reduction in the difference of the refractive indices at the boundary makes it virtually impossible for the average eye to focus a sharp image on to the retina. In a similar manner to the sclera, the cornea ‘yellows’ as it ages.
Conjunctiva
Beginning at the edge of the cornea, the conjunctiva is a membrane that covers the outer part of the sclera and inner surfaces of the eyelids. It serves to prevent dust and other objects from entering the eye and to reduce friction between the eyelids and eyeball by helping to keep it moist. Conjunctivitis, sometimes known as ‘pink eye’, is inflammation or irritation of the conjunctiva.
Iris and pupil
A coloured, muscular, disc-shaped structure, the iris is positioned just in front of the crystalline lens (Figure 4.2). It has a hole in the centre, the pupil, and muscles which run radially and tangentially in a ring around it. When looking in a mirror, it is the coloured portion of the eye that we see. The colour of the iris is determined by, among other things, the amount and ratio of different types of melanin contained within it, either darker brown–black eumelanins or lighter, red–yellow pheomelanins. Melanin is also found in the skin and hair and, in a similar manner, the ratio determines our skin or hair colour. The iris is considered part of the anterior vascular tunic and its rear surface is pigmented, as previously mentioned, a deep purple colour.
Figure 4.2 External photography of the human eye.
The iris controls the amount of light entering the eye which, when fully open, has a diameter of approximately 8 mm in low light levels and around 2 mm in bright conditions. It creates effective apertures from f/11 to f/2, but this only partially accounts for the large range of the lighting conditions under which the eye can operate. Controlled by the central nervous system, the iris has a sphincter muscle about 1 mm thick near the pupil, which constricts it in bright conditions. As light levels drop, radial muscles act to dilate the iris. Testing the reaction of the iris to light and its evenness in both eyes is often used as a basic test for brain function.
The iris is the dividing point for the anterior and posterior chambers located in the front of the eye. The posterior chamber is formed between the rear surface of the iris and the anterior surface of the crystalline lens. It is also filled with aqueous humour, which helps to nourish the crystalline lens.
The texture and patterns exhibited by the iris are believed to be unique and have been exploited in recent years for use in security systems. Though a full description is inappropriate here, most systems rely on a wavelet-based frequency analysis of the patterns and this is covered in more detail in Chapter 28.
Crystalline lens
Known as accommodation, the fine focusing necessary to produce a sharp image is undertaken by the crystalline lens. Constructed primarily of proteins, it is many layered and has a varying refractive index, from approximately 1.40 in the centre to 1.38 in outer layers. Flexible in nature, the diameter of the lens is approximately 9 mm in adults and up to 5 mm thick. The lens is suspended in its own corpuscular bag by zonular fibres behind the iris and is attached to the ciliary body.
The lens is biconvex with approximate radii of curvature of 9 and 6 mm for the anterior and posterior surfaces respectively when relaxed. Therefore, the lens contributes some optical power even when looking at distant objects, but primarily aids focusing of objects closer than about 20 ft. The effective focal length of the combined cornea–eye lens system is approximately 16 mm which, given the overall diameter of the eyeball at 24 mm, some may find surprising. The aqueous and vitreous humours, however, effectively lengthen the optical path between the lens system and the focal plane by reducing refraction at the lens boundaries internal to the eye. The optical power of the lens contributes around a quarter of the total power of the eye. Focusing, or accommodation as it is known, is considered in further detail later in this chapter.
Ciliary body
The ciliary body is a doughnut-shaped piece of tissue that lies at the outer edge of the iris and to which the crystalline lens is attached via the zonular ligaments. Somewhat like the iris, the ciliary body contains radial and tangential muscles which are responsible for adjusting the shape of the lens. The ciliary body is also thought to produce aqueous humour, which diffuses through zonular fibres into the posterior chamber of the eye, through the pupil and into the anterior chamber.
Vitreous cavity and vitreous humour
The vitreous cavity is the main space behind the crystalline lens and in front of the retina. It is filled with vitreous humour, a fluid much the same consistency as egg white, and helps to maintain the shape of the main body of the eyeball. The vitreous humour is predominantly water with a small amount of protein, which imparts its viscous nature and has a refractive index of around 1.337. It is attached to parts of the retina, though ageing generally causes it to liquefy, allowing it to move about. The liquefaction of the vitreous humour and further discontinuities in it may generally be seen as floaters, small spots or lines which move about the visual field as the position of gaze is changed. These are most easily seen on light-diffuse uniform backgrounds. More serious disruption of the vitreous humour can cause tension on the retina and a sensation of flashing lights. If this occurs, or the number of floaters seen increases, it is important to seek professional medical attention as soon as possible to avoid further damage to the retina. Generally, the light-sensitive cells in the retina do not cause pain in response to stimuli. Thus, flashing lights may be the only indication of a problem.
Retina and choroid
The inner surface of the eye starting behind the ciliary body is coated with the retina. The retina is a layer, about 0.5 mm thick, of photosensitive and nerve cells whose task it is to encode the incoming light into electrical signals for the brain. Coating about two-thirds of the inside of the eye, the retina lies on top of the choroid. As previously mentioned, the choroid is considered part of the posterior vascular tunic and primarily provides the blood supply to the retina. Directly above the choroid lies Bruch’s membrane, which separates the retina from the vascular network. Between Bruch’s membrane and the photosensitive cells of the retina lies the retinal pigmented epithelium. This layer helps to exchange waste products and nutrients between the choroid and the photosensitive cells of the retina and further absorb stray light. The posterior surface of the choroid is known as the tapetum and may appear iridescent.
Figure 4.3 A demonstration to find the blind spot. By covering your right eye and looking at the picture of the dog from approximately 20 cm, its owner should disappear. Alternatively, covering the left eye and looking at the owner, the dog will disappear. You may have to adjust the distance that you are to the page.
The detailed structure of the retina is considered later in this chapter.
Optic nerve
The signals generated by the cells in the retina travel towards a light-coloured spot on the retina, known as the optic disc. This is the point at which the optic nerve connects and is located at about 10° from the optical axis on the nasal side. It is additionally known as the blind spot due to the lack of photoreceptors at this point. Approximately 1 million nerve fibres commence their journey from the retina to the visual cortex here, though the pathway from the retina is not direct, as will be examined later. The optic nerve also carries the retinal artery and vein, transporting blood both towards and away from the eye.
Demonstration that the blind spot is insensitive to light is readily achieved. By covering your right eye and looking at the picture of the dog in Figure 4.3 from approximately 20 cm, its owner should disappear. Alternatively, covering the left eye and looking at the owner, the dog will disappear. You may have to adjust the distance that you are to the page.
Structure of the retina
The retina is a light-sensitive layered structure coating the inside of the eye and is considered as part of the nervous, or inner, tunic. It begins at, and is continuous with, the optic nerve and ends just behind the ciliary body at the ora serrata, so named because of its appearance as an irregular margin. A thin, non-imaging, layer continues beyond this point to cover the back of the ciliary body and iris. The retina is considered part of the central nervous system as it derives from the same material as the brain in embryonic development. Five main layers may be identified in the human retina: the receptor, outer plexiform, inner nuclear, inner plexiform and ganglion cell layers. More detailed descriptions, however, often divide the retina into 10 layers, or even more. Figure 4.4 presents a schematic of the basic retinal layers. The first interesting thing to note is that the light-sensitive receptor layer is nearest the choroid. Light must traverse the remaining layers before being detected by two types of cells that exist in the sensitive layer, rods and cones. The rods and cones convert incoming light into electrical signals which then traverse the layers until being transmitted by the axons of the ganglion cells that form the optic nerve. There are approximately 125 million photoreceptors in the retina; however, there are only about 1 million ganglion cells. This may be interpreted as representing compression of the image in the order of 125:1. The significance of this level of compression may be illustrated by comparing an image that has been compressed to a level of 96:1 to its original using the commonly available Joint Photographic Experts Group (JPEG) format (Figure 4.5). Whilst it may be argued that at the time of writing there are better performing compression systems available, JPEG is by far the most common. It should also be noted that higher compression ratios may also be easily achieved by exploiting temporal correlation. Temporal correlation may be thought of as similarity between two or more successive images. Imagine a newsreader in a studio; only the newsreader moves and the background stays the same. Therefore, the background areas exhibit temporal correlation and this information need not be transmitted again.
Figure 4.4 A schematic of the layers of the retina.
Figure 4.5 (A) An image of the centre of Singapore at night. (B) The same image compressed to 96:1 using JPEG compression.
The interaction of the layers in the retina is highly complex and the detailed mechanisms of vision, which involve the organization of the receptors and the way in which signals are generated, organized, processed and transmitted to the visual cortex, are beyond the scope of this book. However, they give rise to a number of visual phenomena which have been extensively studied and have consequences in our understanding and evaluation of imaging systems. In addition there are a number of strong psychovisual effects due to interpretation and processing of the images in the visual cortex and other areas of the brain. Furthermore, upbringing has also been shown to influence the way in which images are understood. A number of generalized effects are given in the following pages and more detailed descriptions are given on colour vision (Chapter 5). For an in-depth description the reader is invited to consult some of the works listed in the Bibliography.
Rods and cones
The primary layer of light-sensitive cells is also known as Jacob’s membrane. Named due to their shape, photosensitive rods and cones are not uniformly distributed throughout the retina and perform synergistic, but differing, functions. Rods are highly sensitive and provide monochromatic vision at low light levels: scotopic vision. Less sensitive by a factor of hundreds, three variations of cones provide colour vision in more abundant levels of light: photopic vision. Mesopic vision occurs when light levels are such that both rods and cones are being used. Scotopic vision exists because cones cannot operate at low light levels and is generally considered to function in an illumination range of approximately 10−6 to 10−2 cd m−2. The lower figure is the threshold at which rods operate and the upper figure that at which cones begin to operate. Because scotopic vision relies mainly on rods it displays poor acuity. Mesopic vision relies on both rod and cone vision from approximately 0.034 to 3.4 cd m−2, or moonlight to twilight. Because cones are not operating optimally it gives poor colour discrimination though slightly better acuity than scotopic vision. For illumination levels above approximately 3.4 cd m−2, cones function optimally and the best colour and acuity vision is rendered.
Figure 4.6 shows the distribution of the rods and cones throughout the retina. It may be seen that there is a very high density of cones at a point on the retina, named the fovea, and relatively few distributed throughout the remainder of the area. Conversely, it may be seen that rods are extremely sparse at the fovea, numbering higher elsewhere in the retina and then dwindling in number towards the periphery of vision. At the blindspot neither rods nor cones exist.
Figure 4.6 The distribution of rods and cones.
The fovea is a point on the optical axis of the eye approximately 0.5–1 mm in diameter (about 2° of the visual field) into which is packed in excess of 50,000 cones in a hexagonal pattern. It appears as a small dip within a yellow circle known as the macula. Rods and cones have sizes in the range of microns, though the cones are slightly thinner in this area and generally have a near one-to-one correspondence with optic nerve fibres at the centre of the fovea. The fovea therefore provides high acuity colour vision in good lighting: photopic vision. The yellow pigment of the macula is know as macula lutea and is thought to reduce chromatic aberration of the eye by absorption of blue light. Acuity drops dramatically towards the periphery of vision and, as a consequence, light must fall on the central 15° or so to provide quality vision. A direct effect of this is that the human visual system scans to build a picture of its environment by constantly changing its position of gaze. These movements are known as saccades and are discussed in more detail later in this chapter.
Though rods outnumber cones by about 20 to 1, approximately 120 million as opposed to 6 million, they provide lower acuity vision in the remainder of the visual field. Spatial resolution is sacrificed increasingly toward the periphery of vision due to increased numbers of rods being ‘wired together’ in order to increase sensitivity and the perception of movement or temporal resolution. Triggering a turn of the head towards the stimulus, the detection of movement in the periphery of the visual field is thought to be important to ready the body for a ‘flight or fight’ response from a would-be predator.
The basic structure of rods and cones is shown in Figure 4.7. Rods, as their name suggests, are approximately cylindrical in shape. The outer light-sensitive part of the rod cell contains hundreds of discs, lamellae, containing the light-sensitive purple pigment rhodopsin, also known as visual purple. On exposure to light the rhodopsin is bleached or broken down, producing an electrical potential and chemical by-products. The electrical potential is the basic building block of the signal that is eventually processed and sent to the brain. Rods are many hundreds of times more sensitive to light than are cones, and this is further enhanced by their being combined in increasingly large clusters, or receptive fields, towards the periphery of the field of view. Therefore, to express it in modern terms, it is not useful to think of a rod or a cone as a single ‘pixel’ in an image. Rather, as will be seen later, the receptive fields to which a number of rods or cones contribute should be thought of as a basic picture element – larger, more sensitive pixels at the edge of the field of view and smaller, higher resolution, but less sensitive pixels at the centre. The increased sensitivity of the rods at the periphery of the field of vision is important to astronomers. The lack of rods at the fovea leads to very poor night vision on the optical axis, so looking directly at an object can often cause it to ‘disappear’. By looking off-axis and using the more light-sensitive part of the retina is it possible to detect objects that are many times fainter.
The spectral sensitivity of the rods may be seen in Figure 4.8. The spectral sensitivity of a system may be thought of as how sensitive a detector is to each wavelength of light considered. If, for example, the sensitivity of a system to green light (550 nm) is 1 and to blue light (450 nm) is 0.5, it will take twice as much blue light to produce the same response as the green, as the system is only half as sensitive. If the spectral sensitivity of a system is zero, it cannot detect that wavelength of light. A related term, spectral responsivity, is the electrical output of a detector compared to the light falling on it with respect to wavelength. Whilst spectral sensitivity may often be expressed as a relative fraction and may be unitless, spectral responsivity is an absolute value with units such as VJ−1 cm−2.
Figure 4.7 The structure of rods and cones.
Figure 4.8 The spectral sensitivity of rods and cones.
Adapted from Dowling (1987)
It is reasonable to expect that because of the rods’ higher sensitivity, they would have a broader spectral sensitivity curve than that of any of the individual cone cells. This, however, is seen not to be the case and the curve for the rods is similar in width to that of any of the cones with a peak response at about 500 nm. This would suggest that the increase in sensitivity is achieved by some other means than simple spectral integration of light. The rods connect to a number of horizontal and bipolar cells contained within the outer plexiform layer of the retina.
Whilst there is only one type of rod, there are three types of cone that have been identified in the human retina, the distribution of which dominates the fovea. In other species these numbers can vary and the proportion of rods to cones, as well as type, can be reflective of an animal’s typical behaviour and habitat. Often denoted short (S), medium (M) and long (L) wavelength cones, they are by no means found in equal numbers. Whilst ‘blue’-sensitive S-cones have the highest sensitivity, they are generally not found in the central fovea and number only 1–2% of the cone total. ‘Red’-sensitive L-cones can outnumber ‘green’ M-cones by as much as 2:1, yielding approximately 64% and 32% respectively. As may be seen in Figure 4.8, red, green and blue are terms that should be used with some caution as the cones’ spectral sensitivities are not narrow-cut and idealized. Rather, they have broad-based curves which overlap greatly and therefore ‘greenish-blueish’ may be a more appropriate name for an M-cone, for example. The peak sensitivities of the S, M and L cones can be seen to be approximately 420, 534 and 564 nm respectively. The integrated peak sensitivity of the cones is close to 555 nm, which corresponds closely to the peak output of the sun.
In Figure 4.7 it can be seen that the structure of the cones is similar to that of the rods, with lamellae contained in the outer segment. Rather than being discs as in the rods, the lamellae are formed as a continuously folded sheet. The lamellae employ three variants of iodopsin – cyanolabe, chlorolabe and erythrolabe – in the S, M and L cones as the photosensitive chemical and are similar to rhodopsin found in the rods. Vitamin A is important in the synthesis and regeneration of photopsins in both rods and cones, and a deficiency can lead to defective vision. Though the cones have a lower absolute sensitivity to light than do the rods, their response time is quicker, and some studies have suggested that under identical stimulus, the signal from a cone can arrive up to a tenth of a second faster than that from a rod. At the most basic level, it is the detection in the differences of the ratios of the signals from the three cones that leads to the sensation of colour. As for rods, cones also connect to horizontal and bipolar cells in the outer plexiform layer. For cones, however, especially in the foveal region, one photoreceptor often connects to one nerve cell, leading to better acuity.
To reiterate, it is wrong to think of the rods and cones as being a single element in an image. There are complex interconnections between the cells formed by other layers in the retina that modify and enhance these fundamental responses to light. A basic description of the remainder of these layers and the cells is given here and the reader is encouraged to also examine the work of some notable authors listed in the Bibliography.
‘Non-imaging’ cell layers
Synapses from the photoreceptors, bipolar cells and connections from horizontal cells are contained within the outer plexiform layer, whereas the inner nuclear layer contains the cell bodies of horizontal, bipolar and amacrine cells. Horizontal cells are large, connecting photoreceptors and bipolar cells that may be some distance apart. Lateral connections in the retina can cause a spread of the signal generated by photoreceptors of up to 1 mm. Horizontal cells alone, however, do not account for this total and are thought to possibly feed information back to the photo-receptors themselves.
Bipolar cells receive signals from rods and cones, in some cases as above via horizontal cells, and in turn they connect to ganglion cells. They may receive signals from several rods, though a cone is more often connected to a single cell. These cells and their inputs are arranged to have ‘on’ and ‘off centre’ receptive fields, which are discussed later in more detail. All photoreceptor signals that are eventually transmitted to ganglion cells are mediated by bipolar cells regardless of the path taken. The receptive field of a bipolar cell is generated by connections to a group of photoreceptors. It is thought that the surround is constructed via connections with horizontal cells.
The inner plexiform layer contains connections between bipolar, amacrine and ganglion cells. Amacrine cells link bipolar cells and ganglion cells; however, they do not connect with photoreceptors directly and carry information laterally across the inner plexiform layer. There are thought to be many dozens of different types, connecting signals from differing areas of the retina. Because of this, amacrine cells are thought to be responsible for some of the more complex processing that occurs in the retina, such as primitive motion detection. Exact functioning of the cells is a topic of current research in the field.
The ganglion cell layer contains the bodies of the ganglion cells and some amacrine cells. The retinal ganglion cells form the outermost layer. Their axons extend across the face of the retina and collect at the optic disc to form the optic nerve, which accounts for approximately a third of all the nerves entering the brain.
Receptive fields
The signal received by a retinal ganglion cell undergoes horizontal spreading and processing by each of the intermediate layers of the retina. Therefore, it may be influenced by any number of photoreceptors in an area around it on the face of the retina, named a receptive field. The receptive fields of retinal ganglion cells generally increase in size towards the periphery of vision, overlap and have complex interactions. Detailed behaviour is beyond the scope of this book and an introductory description of the topic is given here.
On-centre and off-centre form the two fundamental types of receptive field (Figure 4.9). In diffuse lighting conditions a ganglion cell will maintain a steady firing rate of anywhere up to 20 pulses per second. The frequency of the pulses increases for an on-centre receptive field when light strikes the centre portion and decreases when it hits the surround. Therefore, the maximum response for the cell is not achieved when the receptive field is illuminated strongly and evenly, but rather when the centre is illuminated and the surround dark. For an off-centre field the reverse is true, the maximum response being achieved when the surround is illuminated and the centre dark. Therefore, the difference between the signal from the centre and surround is of more importance than the absolute levels themselves. This mechanism within the retina is of great significance as, within reasonable limits, it dictates that the eye responds more readily to changes in contrast than it does to absolute illumination level (see later). This in turn aids its ability to cope with a great range of illumination conditions. Plotting the ideal stimuli for the on- and off-centre fields (Figure 4.9), they may be seen to be very similar to a Laplacian (and a number of other) convolution filters designed to detect edges in digital images. An edge may be thought of as a change in contrast and similarities may be identified between the two cases. Edge detection is discussed further in later chapters.
On-and off-centre receptive fields are equally distributed throughout the field of view. The receptive field of a retinal ganglion cell may be as large as 1 mm. However, this changes markedly over the surface of the retina and significantly in the fovea. As previously indicated, photo-receptors can feed the inputs of a number of bipolar and horizontal cells, which in turn can feed a number of amacrine and, finally, ganglion cells. Within the fovea it is more likely that a single cone will stimulate a single bipolar cell, which in turn will stimulate a single ganglion cell. Towards the edge of the field of view more photoreceptors are connected to each bipolar cell and consequently stimulate ganglion cells in larger numbers. Because of the overlap of the receptive fields it is possible that a single photoreceptor can stimulate a number of ganglion cells and contribute to an on- and off-centre field, even creating inhibitory and stimulatory responses simultaneously.
The size of the receptive field dictates the range of spatial frequencies that it is interested in. Small receptive fields correspond to high spatial frequencies and large receptive fields to lower spatial frequencies. Receptive fields also exist that respond more strongly to select distances in binocular vision. A number of differing retinal ganglion cells have been identified with vastly differing shapes and sizes, and may be shown to connect to differing numbers of photo-receptors, and have differing speeds of response and function. Some cells help to control saccades or the size of the pupil and others exhibit colour preference. Rare, giant photosensitive ganglion cells, numbering only a few thousand, have also been recently identified. These are thought to assist setting of circadian rhythms within the body.
Figure 4.9 On- and off-centre receptive fields.
DARK ADAPTATION
The eye is expected to cope with an incredible range of illumination conditions, up to 12 orders of magnitude, from around 10−6 to 106 cd m−2, yet the retina has an instantaneous contrast range of about 100:1. Contrast may be thought of as the difference or ratio of brightness of two objects. Therefore, in the above, the brightest object that may be seen is 100 times that of the darkest. The eye is much more dependent upon changes in contrast than in absolute illumination level, as seen above, and this very much aids our perception of the world. Imagine being outside on a bright day. The average illumination may be 10,000 lux, yet under the shade of a tree 100 lux, yet we can see perfectly well in either case. Indoors we may look at a television picture, about 100 lux, or underneath a desk, maybe 1 lux, and we can still see perfectly clearly. In both cases, the ratio of the darkest to brightest thing is 100 and it is the eye that has shifted the scale and adapted to the overall brightness. If the eye did not adapt in the above example and was only dependent on the absolute levels of light, when we arrived indoors we would not be able to see, as nothing would be as bright as our darkest object outside.
Change in size of the iris accounts for some fast and temporary adaptation. It is the adjustment of the chemistry, however, particularly the concentration of unbleached photopsins, that yields the full range over which the eye can operate. When one moves from a brightly lit environment to a dark or dimly lit room, it immediately appears to be dark. After 30 minutes the eye adjusts to the conditions. Chemical adaptation of the eye is well documented and plotting luminance level versus time, as shown in Figure 4.10, exhibits a distorted curve shape. The initial increase in sensitivity is due to cones adapting until the point at which they cannot adjust any further, causing a reduction in the gradient of the adaptation curve. After this point the increase in sensitivity is due only to the rods. There is a further influence on this process by neural effects. A large proportion of dark adaptation takes place within the first 30 minutes, though a slight improvement beyond this can be detected for over an hour. Poor health may also affect the ability of the eye to adapt. Light adaptation is the reverse process with the same mechanism, but is generally accomplished within approximately 5 minutes.
Figure 4.10 Dark adaptation of rods and cones.
The integrated peak sensitivity of cones is close to 555 nm, which corresponds to the peak output of the sun, whereas that of rods is 500 nm (Figure 4.8). This difference causes a shift towards the blue end of the spectrum when using scotopic vision, the Purkinje shift, and an associated decrease in red sensitivity (Figure 4.11). This can cause the brightness of colours with a high red content to appear darker than might be expected when using night vision. This effect should not be confused with metamerism, which is discussed in Chapter 5. Because of the length of time it takes to become fully dark adapted, the reduction in red sensitivity is exploited by those who wish to maintain their night vision, such as astronomers. By making illumination, display panels or other controls red it is possible to read information using the foveal red-sensitive L-cones, whilst causing only a partial reduction in night vision, if at all. Excessive exposure to a bright stimulus will cause desensitization of that area of the retina in the field of view, due to localized bleaching of photopsins. When the stimulus is removed an after-image remains as the retina is no longer sensitive to low light levels. The after-image remains until the local sensitivity is returned to normal. Confirmation that after-images are a retinal, rather than a neural, effect is readily available as it is possible to create them in a single eye by closing the other. When the other eye is used to view the scene the after-image disappears.
Figure 4.11 Photopic and scotopic luminous efficiency of the human eye. The Purkinje shift is seen as the shift in wavelength of the peak sensitivity.
ELEMENTARY COLOUR VISION
Young–Helmholtz theory of colour vision
As detailed in Chapter 1, the splitting and subsequent recombination of light suggested that the human eye might possess three types of colour sensitivity, to blue, green and red light respectively. This triple-sensitivity theory is called the Younge-Helmholtz theory of colour vision. It provides a fairly simple explanation for the production of any colour from appropriate proportions of these primaries. This type of colour mixing is applied in cathode ray tube displays which have red, green and blue light-emitting phosphors in their faceplates, whereas subtractive colour mixtures are applied in most colour photographic materials, colour hard-copy output devices and liquid crystal displays. Subtractive colour mixing involves the overlaying of cyan, magenta and yellow colorants.
Opponent theory of colour vision
The trichromatic theory of colour vision cannot account for some observations. We do not see blueish-yellows nor do we see reddish-greens. Further if we look at the after-image of a red object it is often green and if we look at the afterimage of a blue object it is often yellow. This premise formed the basis of the opponent theory of colour vision: that the neural processing in the visual pathway exhibited red versus green, blue versus yellow and black versus white opponency. Electrophysiological studies appear to support the theory with the discovery of redegreen and bluee yellow ganglion cells. Refer to Chapter 5 for more detail on the subject.
COLOUR ANOMALOUS VISION
Colour anomalous vision, or colour blindness as it is commonly known, is brought about by not being able to detect the ratios of the proportions of the spectrum sufficiently well and usually results in the inability to distinguish certain groups of colours, rather than monochromatic vision as the name suggests. It generally occurs when one of the photopsins cannot be synthesized. Most colour blindness is hereditary, carried by the X chromosomes, and therefore affects many more males than females. Up to 8% of males are affected in some populations whereas less than 0.5% of females are affected. Completely monochromatic vision, achromatopsia, occurs in less than 1% of cases. There are no treatments for colour blindness and it can cause some difficulties such as reading maps with colour legends, traffic lights, colour-based chemical tests, looking at the ripeness of food, selecting clothes or interpretation of coloured graphs. A difficulty distinguishing red and green colours is by far the most common symptom of colour blindness, occurring in 99% of cases, consisting of protans who are red weak and deutans who are green weak.
People with regular colour vision are known as trichromats and their vision as trichromacy (Chapter 5). Anomalous trichromacy occurs when one of the three types of cone does not operate as expected but still functions to some degree. A more severe condition occurs when one of the cone types ceases to function at all, dichromacy.
Anomalous trichromacy may be divided into protanomaly and deuteranomaly. Protanomaly is red-weak vision and the colour’s saturation and brightness suffers as a result. Considering opponent theory, this in turn affects the hue, saturation and brightness of all colours that may be thought of as containing a proportion of redegreen opponency: reds, oranges, yellows, greens, purples. Protanomaly affects approximately 1% of males. Deuteranomaly is green-weak vision and causes similar discrimination problems as protanomaly in approximately 5% of cases, again affecting colours with redegreen opponency, though with greens appearing pale and unsaturated.
Dichromacy affects about 2% of males and can be further divided into protanopia and deuteranopia. Whilst anomalous trichromats can readily function, these conditions are usually more severe. Rather than having a slight effect on reds or greens they can generally see no difference between the hues. Protanopia, a severe red deficiency, affects 1% of males. Reds, oranges and yellows appear dark and cannot be distinguished from greens. Deuteranopia, severe green deficiency, also affecting around 1% of males, is similar but does not cause abnormal dimming. Blueeyellow colour blindness also exists, though is very rare.
Tests for colour blindness are readily available and easily administered. For diagnostic purposes lighting conditions should be arranged so that they simulate daylight at a comfortable level of illumination. If other lighting arrangements are used results will generally only be valid for those conditions. The Farnsworth–Munsell 100 hue test consists of four trays with a total of 85 caps exhibiting slight changes of hue. Users attempt to arrange the caps in the order of their hue. Errors in the cap positions are logged and used to diagnose the quality of the subject’s colour vision. The FarnswortheMunsell 100 hue test can detect all types of colour vision and the 85 hues form a perfect hue circle.
A pseudo-isochromatic plate appears as an image containing random dots of differing colour. Hidden within the plate is a number or pathway of a slightly different hue. People with colour anomalous vision are unable to read the hidden information. An example of this is the Ishihara plate test, the full version containing 38 plates and commonly used for detecting redegreen blindness. American Optical Plates may further be used to grade severity of colour blindness. Modified plate tests employing pictures rather than numbers may be used to test people who have difficultly communicating or reading, such as preschool children.
MOVEMENT AND FOCUSING
Focusing and correction of eyesight
Focusing of the eye is known as accommodation. The crystalline lens is suspended behind the iris by the zonules of Zinn (Figure 4.1). The zonules are ligaments, made of collagen, which attach to the circular ciliary muscle. An out-of-focus retinal image triggers the parasympathetic system, which contracts and relaxes the ciliary muscle. As the ciliary muscle is relaxed the zonules become taught, placing tension on the crystalline lens and it is flattened. When the ciliary muscle tightens the zonules relax and the lens becomes rounded. To maintain focus on distance objects, the curvature of the lens is reduced, the lens is flattened and the focal length is increased. Conversely, to focus on close objects, the curvature is increased, the lens fattened and the focal length reduced (Figure 4.12). Focusing in other animals can include moving the lens rather than changing its optical power in the manner above. As the eye ages the crystalline lens becomes thicker and stiffer because its proteins continue to grow. This causes it to harden, diminishing its ability to change shape and therefore focus. Known as presbyopia, this generally starts to occur after age 40 and makes it more difficult to focus at a near distances. Weakening of the ciliary muscle adds to the effects of the condition.
If the cornea has too little curvature (flatter than it needs to be) the optical power of the crystalline lens is unable to compensate for this and images are brought to a focus behind the retina. Hyperopia, or far-sightedness as it is commonly known, occurs in approximately 1 in 4 people and causes near objects to be out of focus. The condition may also occur if the eyeball is too short. Adding a positive power lens in front of the eye can correct this (Figure 4.12a). Myopia, or near-sightedness as it is known, is the complementary condition, again affecting approximately 1 in 4 people. Images are focused in front of the retina either because the power of the cornea is too great or because the eyeball is too long (axial myopia). The addition of a negative power lens to the front of the eye may correct this.
Figure 4.12 (a) Focusing on a distance object, the lens flattens. (b) The lens becomes more rounded when focusing on near objects. (c) Hyperopia, or commonly far-sightedness, may be corrected by adding a positive power lens in front of the eye. (d) Myopia, or nearsightedness as it is known, may be corrected by the addition of a negative power lens to the front of the eye.
Figure 4.13 (a) The effect of astigmatism of the focal length of the lens in tangential and sagittal planes. (b) The effect of astigmatism on vision. The bars become blurred at a range of angles.
Astigmatism is an unsymmetrical curvature of the cornea or crystalline lens. The affected surface curves more in one direction than another, somewhat similar to a rugby ball. If a number of cross-sections are taken each will exhibit a different optical power or focal length (Figure 4.13a). The effect of this on focusing is interesting and causes those affected to be able to focus on a structure in a single direction more strongly. For example, it is possible that somebody may be able to focus on horizontal lines but not vertical lines at the same time (Figure 4.13b). Astigmatism, occurring in isolation, may be corrected using a cylindrical lens.
Movement
Because of the change in resolution across the field of view of the eye, movement and scanning are very important to build up a picture of the scene around us and proper perception of the world. Though it is generally not noticed, the eye is moving constantly to direct the small, high acuity area of the fovea to objects of interest. Additionally, the eye needs to be able to compensate for movement of the head in order to be able to maintain direction of gaze and to follow objects that would otherwise move too quickly across the retina. The maximum speed is just a few degrees per second before the brain fails to recognize moving images. Each eye is controlled by six muscles, which are attached to the sclera.
Eye movements may be divided into a number of differing types: smooth pursuit, saccades, vestibular-ocular reflex, opto-kinetic reflex and vergence. Smooth pursuit occurs when an object is followed at speeds of up to 100° per second, whereas a saccade is a rapid movement of the eye to a different part of a scene. Saccades may reach speeds of up to 1000° per second and are ballistic: they may be initiated voluntarily but cannot be changed once started. They are used primarily for fixation but imaging is suppressed during movement, as is the perception of the movement (saccadic masking). A micro-saccade is an involuntary movement similar in nature to a saccade but occurring over a much smaller visual angle, approximately 0.2°. Their exact purpose is still a topic of research, though it has been suggested that because the eye is continually looking for contrast they may be used to refresh the retinal image or to stimulate different receptive fields. Direction of gaze, as the head is moved, is maintained by the vestibular-ocular reflex and the balance organs in the ears are used as a feedback mechanism. The effectiveness of this method may be demonstrated when reading a book. Moving one’s head from side to side it is possible to continue reading a motionless book, whereas keeping one’s head still and moving the book it is not possible to continue reading beyond the most modest of speeds. Saccades and smooth pursuit are combined by the opto-kinetic reflex. The eyes follow the direction of movement of an object and a saccade returns the eye to the starting position. This is commonly experienced as a passenger in a car or on a fairground ride. Human vision is a binocular system and if the images from each eye did not overlap double vision would occur. For an object to be brought to the centre of the field of view of both eyes, it is necessary to turn the eyes inwards slightly. This is known as vergence and without it binocular vision would be substandard. It is also possible to roll the eyes about the optical axis and this is generally dependent on the angle of the head.
THE VISUAL PATHWAY
The response of the human visual system to stimulus is generally so good that we do not consider it on a regular basis. Yet the process and the brain are incredibly complex. The brain has approximately 1012 cells. The visual pathway starts with the signals generated in the 125 million or so photoreceptors that line the back of the eye and concludes with a perception of that scene, further comprising all interconnecting points. Brain function is an active area of research and a great deal of its workings are unknown; however, considerable efforts have been devoted to the visual pathway.
The visual pathway may be considered as largely serial, although there are some parallel branches (Figure 4.14). The signal from photoreceptor cells is subject to processing by each of the retinal layers as described previously. The result is transmitted by axons of ganglion cells which form the optic nerve, containing approximately 1 million fibres, and travels to the optic chiasm.
Each half of the visual field is generally referred to as the temporal and nasal fields respectively. Each eye generates largely overlapping information. Fibres from the nasal field in each eye cross at the optic chiasm, causing the left and right halves of the scene to be joined. The optic chiasm then continues in two pathways known as optic tracts. The right visual field heads towards the left brain and the left visual field heads towards the right brain. The overlapping visual fields are important for stereo vision and the optic chiasm is therefore crucial to facilitate this.
The optic tracts head towards two peanut-sized areas within the thalamus known as the lateral geniculate nucleus (LGN). The exact functioning of the LGN is not fully known. It is a layered structure and has six layers of cell bodies, some of which interconnect, as well as others sending pulses directly to the visual cortex. It is shown to have receptive fields as in the retina and is thought to preprocess information and introduce coding efficiencies by cancelling redundancy. It may also help to coordinate the attention of the visual system to important events from other sensory stimuli. For example, a sound from the right may draw the gaze of the eyes. LGN axons fan out through the white matter of the brain via the optic radiations and ultimately travel to the primary visual cortex. It is also known that the LGN has feedback connections from the primary visual cortex. Signals are not only distributed by the optic nerve and chiasm to the LGN; information is also distributed to additional areas of the brain to control processes such as the convergence of the eyes or generate saccadic movements.
Figure 4.14 The visual pathway.
Visual cortex
The visual cortex, one of the largest brain structures, is located at the rear of the head. It consists of the primary visual cortex, also known as the striate cortex, and extrastriate cortex. The term striate derives from its layered appearance and structure. The primary visual cortex is further designated area V1, and the extrastriate cortical areas V2, V3, V4 and V5.
The primary visual cortex receives information directly from the lateral geniculate nucleus, after which there are thought to be two main processing pathways, the ventral and dorsal streams. Though still a subject of current research, these are often referred to as the ‘what’ and ‘where’ or ‘how’ pathways respectively. The ventral stream is thought to be responsible for recognition and memory, traversing areas V1, V2 and V4. From V4, information travels to the inferior temporal lobe. The dorsal stream is thought to be associated with the position and movement of objects and travels towards the inferior parietal lobe after traversing V1, V2, V3 and V5.
The field of view is mapped to the area of V1, preserving spatial information. Cortical magnification, however, maps the central portion, namely the fovea, to a much larger proportion, approximately half. Receptive fields have been shown to exist throughout the visual cortex, as in the lateral geniculate nucleus and the retina, though are more sophisticated than in either.
BINOCULAR VISION
The perception of depth is incredibly useful for performing many everyday tasks – driving, walking, picking up objects. It enables the judgement of distances and the relative speeds of things moving towards or away from us. The primary mechanism of depth perception is stereopsis. Images from both eyes show slightly different views of the same scene as they are taken from different angles. The brain is able to interpret this as a single view through analysis of the images. To be able to accomplish this, the field of view from each eye must overlap.
Besides stereopsis, there are many other effects that give binocular clues to the visual system. Overlapping objects suggest that one may be behind another. Also, if size remains constant they will appear larger as they approach the eye. Because of perspective, it is generally also expected that stimuli will appear higher in the scene and subtend a smaller visual angle. Colour will also affect apparent depth as atmospheric haze tends to desaturate and lowers contrast. Desaturation may be thought of as a loss of colour intensity. More subtly, textures also repeat with increasing frequency. The angle that the eye needs to turn in to bring an object to the centre of the field of view in each eye, convergence, decreases with distance. Finally, if movement is involved, movement parallax may be utilized. This appears as closer objects moving more quickly than those further away. This is because the extent of a scene included in the field of view increases with range for a constant visual angle. Conversely, moving the head will cause objects that are near to apparently move more. Measuring parallax, by recording the associated position in the image as an observer’s point of view is changed, allows distance to be calculated by triangulation. The technique may be used for rangefinding or measuring the distance of stars, for example.
The use of two eyes introduces other effects. Because of binocular summation, the threshold for detecting a stimulus using both eyes is lower than that for one. Interaction of the eyes leads pupil diameters to be the same in both, even if one is closed and, further, if the open eye is focused on an object, the accommodation of the other will be the same.
Depth from stereopsis arises from a form of parallax. Rather than changing the point of the view, the parallax is generated by the eyes being separated by a distance. The slight difference in positions of the image on each retina generates retinal disparity, the depth clue. Retinal disparity may also be called horizontal disparity or binocular disparity. To be able to detect retinal disparity the same point in both retinal images needs to be matched, or correlated. This leads to a correspondence problem – any point in one retinal image can potentially be matched with a number of points in the other. It is thought that the visual system overcomes this by placing receptive fields at slightly different horizontal positions throughout the visual cortex. This essentially configures matching left and right eye receptive fields to respond more strongly when stimuli exhibiting the correct horizontal disparity are presented. Fusion of images from both eyes to provide a single scene is considered separate to depth perception. This is known via disorders of the visual system that may destroy stereopsis but spare fusion. Predatory animals often favour stereo vision over a large field of view to be able to locate and chase prey by placing the eyes for large overlapping fields. Prey, conversely, often favour a larger total field of regard and thus avoid being caught. The behaviour of the human visual system is not completely encoded by the genes; a proportion of it is learnt through experience as infants. A lack of visual stimulus during development can lead to poor depth perception.
PERFORMANCE OF THE EYE
Luminance discrimination
Discrimination of luminance (changes in luminosity – lightness of an object or brightness of a light source) is governed by the level. As luminance increases, larger changes in luminance are needed to perceive a just noticeable difference, as shown in Figure 4.15. This is known as the Weber – Fechner Law and over a fairly large luminance range the ratio of the change in luminance (ΔL) to the luminance (L) is a constant of around 0.01 under optimum viewing conditions:
The ratio of light intensities is the significant feature of our perception. Figure 4.16 gives a visual indication of reflected light intensities in which each step increases approximately by an equal ratio of 1, 2, 4, 8, i.e. a logarithmic scale. Stellar magnitudes are expressed logarithmically for this reason. Hipparchus ranked the brightness of the stars that he could see from 1, the brightest, to 6, the faintest, in ancient Greece in about 150 BC. The system, though extended and formalized, is used to the present day. The logarithmic response of the eye creates the need for gamma correction in a number of imaging systems, as is discussed in later chapters.
Figure 4.15 Response vs. intensity for the human eye.
Figure 4.16 Equal steps in lightness, each step differing by an equal ratio.
Contrast sensitivity function
The human visual system’s ability to discriminate fine detail has been determined in terms of its contrast sensitivity function (CSF). The CSF is defined as the threshold response to contrast where contrast (or modulation) is the difference between the minimum (Lmin) and maximum (Lmax) luminances of the stimulus divided by their sum:
A typical CSF for luminance shown by the HVS is shown in Figure 4.17. It is usually measured by presenting sine waves of various frequencies to the user and varying the contrast until the pattern can just be detected. Because of the distribution of rods and cones in the retina the CSFs for the colour channels (see Chapter 5) are different from those shown in Figure 4.17, with lower peaks and cut-off frequencies. For luminance the HVS has a peak spatial contrast sensitivity at around 5 cycles per visual degree and tends to zero at around 50 cycles per visual degree. Because of the lower acuity exhibited by rods in scotopic vision, it would be expected that the CSF curve would change with respect to illumination. Indeed, this is the case and overall performance drops as a function of illumination. Further models of the performance of the eye may be found in Chapter 5.
Figure 4.17 Spatial contrast sensitivity function for the human visual system.
Visual acuity
Whilst the CSF for the eye expresses its ability to resolve individual spatial frequencies it gives no intuitive information about its ability to perform basic tasks or performance against the average population. Visual acuity is an indication of the resolving power of the eye. The height of a test chart letter is arranged to subtend an angle of 5 minutes when viewed at a distance of 20 ft. This equates to a height of approximately 8.9 mm and normal vision. The height of the letter is chosen according to the Raleigh Criterion (Chapter 2), such that each line of a capital letter ‘E’ should be resolved correctly. The distance is chosen so that the eye is at optical infinity. Visual acuity is then written as a fraction with the numerator being the distance of the test and the denominator an indication of the quality of vision. Normal vision is therefore written as 20/20. Good vision may be written as 20/10; the subject can read a letter at 20 ft that a normal person can only read at 10 ft. An example of poor vision may be 20/30, where the chart could be read at 30 ft by somebody with regular vision.
ANIMAL VISION
The vision of animals exhibits some remarkable variation and the diversity inspires many ideas in imaging. Most of their visual systems have evolved to best cope with their environment and behaviour. Nocturnal animals, for example, usually have highly dilated pupils to let in as much light as possible as well as proportionally larger corneas. As mentioned previously, predators often have large forward-facing eyes and utilize stereopsis. Some of the simplest eyes, ocelli, are found in animals such as snails and only have the ability to distinguish between light and dark.
A number of common misconceptions exist regarding the vision of various animals, and the often misquoted monochrome vision of the dog is an example of this. Whilst the canine retina is comprised mostly of rods, two types of cone do exist in low numbers, making them dichromates, with 10–20% cones in the central retina. The cones have peak sensitivity at approximately 429 and 555 nm.
The predominance of rods leads the canine eye to respond well in low light and this is further enhanced by a large cornea and the tapetum lucidum, a reflective layer that exists in the choroid, reflecting any unabsorbed light back towards the photoreceptors. The tapetum can shift the wavelength of light, via fluorescence, towards the peak sensitivies of the rods. Instead of a fovea, the canine retina contains an elongated area named the visual streak. Parallel to the ground, it is located above the optic nerve. Canine visual acuity is around a third that of a human, typically from 20/50 to 20/100, and the optic nerve reflects this, consisting of less than 170,000 fibres. The feline visual system is also adapted to low light vision, though weakly trichromatic, and a lower visual acuity, approximately 20/100. The tapetum can appear iridescent in both species, lending its name to the retro-reflectors marking the centre of roads, ‘cats’-eyes’. The feline tapetum can reflect over 100 times more light than a human eye. Both canine and feline eyes have elliptical pupils, allowing them to close almost completely.
It is the airecornea interface that provides the majority of the optical power in the human eye and it is the reduction in the difference of the refractive indices that renders the eye unable to focus when underwater. To overcome this problem some animals, such as crocodiles, have additional refractive eyelids in order to be able to see both in and out of the water. The optical power of the eye of some diving birds can change by up to 50 dioptres as opposed to around 16 for a human. Avian eyes exhibit further remarkable diversity. The hawk relies upon eyesight for hunting and this is reflected in the performance, with visual acuity being estimated at 20/2. It is thought that it can detect prey from a height of about a mile. Raptor eyes contain rods just like humans but can have up to five types of cones enhancing colour discrimination, and more tightly packed photoreceptors. Oil droplets in the cones act as filters to select appropriate wavelengths. A large majority of avian eyes, and many animals, also contain nictitating membranes, an extra eyelid which is used to wipe dust and dirt from the eye regularly. Unlike the outer eyelid these generally move horizontally.
Figure 4.18 Ommatidium of the compound eye.
When looking at the sky light is partially polarized (see Chapter 2) and this is reflected in the recently discovered ability of pigeons to be able to detect the plane of polarization. Whilst in pigeons the ability may aid homing, in cephalopods polarization has been developed for hunting. Giant squid, for example, operate in the low light levels of the deeper ocean and as a result their eyes have developed very large apertures of around 10 inches and better than f/1. It is thought that polarization helps them to see transparent or semi-transparent prey. Their visual acuity is estimated as being twice that of humans, though many believe that they are also colour blind as most only have a single visual pigment. Some cephalopods can also display polarized patterns on their bodies using iridophores, such as cuttlefish. Because seawater heavily filters light, discrimination of colour becomes more difficult at even modest depths; however, polarization remains unaffected.
Insects generally rely on compound eyes, which consist of many repeating units, ommatidium (see Figure 4.18). Ommatidia typically number thousands, with each one directed towards a narrow field of view. Each has an individual lens focusing light on to photoreceptors. Acuity varies amongst insects, though it is thought to be approximately one-sixtieth that of humans for a honeybee. The compound eye has a wide field of view, is relatively simple and is thought to be excellent for detecting motion. It has been observed that insects respond more readily to moving objects. Some compound eyes are able to distinguish colour. The bee, for example, has photosensitive cells with peak absorptions of 344, 436 and 544 nm. Flowers often exhibit patterns in the ultraviolet region of the spectrum which are invisible to the human eye and this is thought to explain the sensitivity range.
BIBLIOGRAPHY
Barten, P.G.J., 2000. Contrast Sensitivity of the Human Eye and Its Effects on Image Quality. SPIE Press, USA.
Bruce, V., Green, P.R., Georgeson, M.A., 1996. Visual Perception: Physiology, Psychology and Ecology. Psychology Press, Hove, UK.
Dowling, J.E., 1987. The Retina: An Approachable Part of the Brain. Belknap Press of the Harvard University Press, Cambridge, MA, USA.
Falk, D., Stork, D., 1986. Seeing the Light: Optics in Nature, Photography, Color Vision and Holography. Harper & Row, New York, USA.
Gordon, S., 2000. The Aging Eye. Simon & Schuster, New York, USA.
Hubel, D.H., 1989. Eye, Brain and Vision. Scientific American Library, New York, USA.
Jackson, R., MacDonald, L., Freeman, K., 1994. Computer Generated Colour. Wiley, Chichester, UK.
Marr, D., 1982. Vision: A Computational Investigation into the Human Representation and Processing of Visual Information. W.H. Freeman, San Francisco, UK.
McDonald, R. (Ed.), 1987. Colour Physics for Industry. Society of Dyers and Colorists, Bradford, UK.
Ray, S., 1994. Applied Photographic Optics, second ed. Focal Press, London, UK.
Weizer, J.S., Fekrat, S. (Eds.), 2006. All About Your Eyes. Duke University Press, Durham, USA.