An algorithm for segmenting juxtaposed objects

This work is an attempt to develop an algorithm for segmentation of juxtaposed objects arising in the recognition of certain scenes. The approach is based on matrix manipulation. The algorithm is implemented in two parts. The first part is to find the significant points of concavity. In the second part, the direction of the paths of segmentation is found and the paths are grown into the pattern until complete segmentation is achieved. The algorithm is illustrated by considering examples from biomedical imagery.