Develeping of Bridging Mode and Global Image Matching

This paper describes three respects in which one-dimensional single image segment bridging mode method may be extended.(1) from one-dimensional image segment to two-dimenional segment for increasing the amount of information int he matching unit,(2) from esingle-feature-oriented to image-segment-oriented matching,and(3) from single image segment matching to global image matching so as to increase the consistency of the entire image.The last respect is concerned with the combination of the feature bridging mode me thod with dynamic programming to reduce the problem of global image matching into one of finding the optimal path.In order to deal with the problem of "occlusion" or "noise feature",the paper introduces the concept of fictitious state,which allows the dynamic programming method to bridge over one or more stages.