Determination of zero-crossings for stereo image matching

The problem of stereo image matching requires compact/ faithful descriptions of a pair of two-dimensional images to compute the depth to points in a three-dimensional scene. Zero-crossings/ corresponding to edges in the underlying image/ prove to be these "primal" descriptors. The microcomputer poses fundamental constraints which confine implementation of an effective solution to this problem. An algorithm is developed and tested at two levels of microcomputer based image processing systems for the computation of one- and two-dimensional zero-crossing descriptions. Multi-scale analysis provides the key for insight into the validity of a microcomputer approach to the problem.