Analysis of Cluster based Self Organization Map (SOM) Algorithm in Various Color Spaces Models

Image segmentation means separation process that can divide the original image into smaller area with similar attributes. In this proposed system, input images are taken from the Berkley Image Segmentation Database (BSD) for color image segmentation. Various color space of images such as RGB, HSV and L*A*B* are used for the segmentation process. Due to effect of the color conversion function reduce the input images size is not flexible, Image J software is used to get the same size of images for different color space. The subjective and objective measured is applied to analyze the color images. Then the cluster based self-organization map (SOM) is applied to produce a low-dimensional input space of the training samples. SOM method develops the ratio of color images similarity and spatial relationship of objects within an image. In this system, the features of color similarity in the image is first segmented into cluster regions and the resulting regions are treated by computing the spatial distance between any two cluster regions and then the labeling process is made by SOM