Improved active contour based object segmentation using color informations

Abstract

This paper proposes a novel method to segment objects based on color informations. The paper analyses the accuracy limitation of the previous methods using grayscale informations. Based on these analyses, the paper improves the object boundary detection by expanding energy calculation equations for color information channels. The algorithm uses all three RGB color channels to increase the accuracy. Simulation results on the large image data set verify the proposed algorithm's efficiency.