A Chromosome Segmentation Method Based on Corner Detection and Watershed Algorithm

A Chromosome Segmentation Method Based on Corner Detection and Watershed Algorithm

연구 분야: Safety

학회: Computer Graphics International Conference

Karyotype analysis is an effective tool for chromosome disease diagnosis, and the number and morphological characteristics of chromosomes can be medically analyzed and described by image processing technology. Chromosome image segmentation is the basis of karyotype analysis. Chromosome images have the characteristics of high adhesion, overlapping and nesting, which is a difficult problem in chromosome image segmentation at present. In order to effectively solve the problem of chromosome adhesion or overlap, this paper innovatively applies watershed algorithm based on gray difference transformation and corner detection to chromosome image segmentation. The algorithm uses gray difference transformation in preprocessing to reduce the phenomenon of image over-segmentation caused by watershed algorithm and separate lightly adhered chromosomes. For overlapping chromosomes, corner detection is used to find the best corner of chromosome segmentation, and then the overlapping chromosomes are separated. Through experiments on 100 chromosome images, the accuracy of chromosome segmentation is 96.2%.