The second big category of advanced image and video processing methods is dealing with stereoscopic images. Stereoscopic images are usually shot either by two normal cameras positioned in parallel and only a few centimeters apart, or by stereoscopic cameras with two lenses and separate image sensors for each one. Either way, the lenses have a distance from each other that resembles the distance between the human eyes, and allows the camera(s) to shoot images that can be fused to simulate three-dimensional vision.

The need for the two fields of view leads to a subsequent doubling of the frame rate and storage space needs. This is because instead of showing just one image, modern 3-D televisions must display two images, one for the left and one for the right eye. Similarly, to store a 3-D video, we need to have double the space compared to a normal 2-D video.

Image registration is again the most significant aspect of 3-D image and video processing. The only way to correctly display the left and right frames is to know the exact correspondence of points from one to the other. The correspondence becomes easier when the camera(s) used are placed very carefully in terms of alignment and distance from each other.

For our first example, we will demonstrate a rather old-fashioned way to make a 3-D video in MATLAB. By old-fashioned, we mean that we will be mixing both the left and right videos into one, with the red color containing the right channel information and the cyan (green and blue) carrying the left channel information. This way, our 3-D video can be watched with an ordinary pair of red-cyan 3-D glasses.

We will assume that you have done the following: