Image Segmentation Utilizing Color-Space Feature

Abstract

In this paper, the authors introduced a stochastic model for color images. Utilizing this model, they proposed a new method for color image segmentation. The proposed method consists of three stages; the first stage considers the red, green, and blue color component of the image as a gray image. One of the known gray image Thresholding algorithm is applied on the three color components. The second stage segments the image based on the results of first stage. This stage produces eight color segments. The third stage identifies the segments through color-space correlation. Color-space correlation algorithm assumes that a set of pixels are considered to belong to one region if and only if they belong to the same color cluster and all connected using neighborhood filters. The last stage may produce very small segments. These small segments are merged with their closed neighbors based on color features. Finally, Conducted experiments achieved perceptually accepted segments and compare favorably to other segmentation methods.