Abstract: The classical image threshold segmentation algorithm converts the color image to grayscale image before segmentation. On the base of analysis upon the characters of RGB color space, the article proposes an image segmentation algorithm based on RGB color space which uses the new criteria to replace the tetrahedral with the cube in color space and segment the color image. The analysis and experimental results indicate that the improved criteria can overcame the wrong which caused by gray degrees conversion and color information lose. As classical algorithm, it is fast and easy to segment the color image. It is suitable to segment the image, which has black target, and can be used to other colors.

Key words: image segmentation, color image, RGB color space