Segmentation of colloidal crystal images by local histogram analysis

The segmentation algorithm given herein is a development of a study of hard-sphere colloidal crystallization undertaken by the authors. The motivation for the development of the algorithm was the lack of a suitable commercially-available program. The desired algorithm is one capable of accurately segmenting a large set of digitized images with a sparse population of blobs varying in contrast to the local background. Furthermore, the program would also need to segment accurately images with backgrounds which varied locally and temporally. The algorithm used here appears to provide adequate results with stable sensitivity to low and high contrast crystal images when implemented recursively