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.

We notice that each kind of image has its own feature and these features can recognize by human brain easily. The difference between person and computer is that person can notice the major interest object of a picture and ignore others. We propose a new segmentation method of satellite images based on fuzzy c-means clustering (FCM) to extract an object, cloud. Our method only promise extracts the cloud from a satellite image clearly just like the weather expert only notice the cloud. We extract R, G, and B components of a true color satellite image and convert a true color satellite image to a 256 gray image, from one image get four images. The ANOVA algorithm was used to analysis the cloud images and its results shown that only the discrepancy between the group of “R-components” and the group of “Gray images” are not significance. In other word, we can use the “R-components” to extract the cloud from a satellite image, instead of the converted gray image, and then the operation time could be shortened.