There are three distinct and widely accepted stages to a stereo image processing algorithm :

(a)

The image features in each image of the stereo pair must be detected and recorded.

(b)

The corresponding image features must be matched across the stereo pair.

(c)

Finally disparity measures must be recorded for the amount of shift each image feature has undergone from one image to the other. This provides the values for constructing the depth map of the cyclopean image, which is the result of combining the left and right images.

• Grey level correlation
• Feature detection
• The stereo correspondence problem
• Random dot stereograms