As one of the most valuable fish, zebrafish has tremendous potential applications in disease modeling and medicine invention. The aim of this paper is to propose an advanced method to implement the automatic detection and recognition of somites of zebrafish.First, various kinds of image segmentation methods are discussed such as threshold segmentation, edge detection, seed growth method, outline taking and watershed method. Base on the algorithm we mentioned above, a novel method is proposed to realize the image segmentation of zebrafish, which is fast and robust compared to other existed algorithms. Then by using square standard, the whole main part of zebrafish is taken out. Making a rotate towards the image is very important. Thus the somites are projected into the horizontal axis and we begin to take the somites. In the end, a new method is proposed to take the somites, which is based on average distance standard .The experiments show that the algorithm is effective towards the zebrafish image. |