Color Image Segmentation of Live Grouper Fish with Complex Background in Seawater

摘要

The color live fish image segmentation is a important procedure of the understanding fish behavior. We have introduced an simple segmentation method of live Grouper Fish color images with seawater background and presented a segmentation framework to extract the whole fish image from the complex background of seawater. Firstly, we took true color pictures of live Grouper fish in seawater using waterproof camera and save these pictures files as RGB format files, called True-color Images. Secondly, we extracted R, G and B planes of a true color Grouper fish image, painted and compared their histograms of R, G and B planes. Thirdly, we segmented these RGB images and the R, G and B planes of a true color Grouper fish image with the k-means clustering algorithm, using the kmeans() function which is packaged by the Clustering Analysis ToolBox of Matlab 2012(a). Finally, we analyzed the relationships between these histograms and segmented images, and then got a conclusion is that: using the B plane of these RGB images as Input-matrix to do clustering segmentation algorithm by the kmeans() function of Matlab Clustering ToolBox, can got a fulfilling segmentation results.