A Level Set Method for Color Image Segmentation Based on Bayesian Classifier
Jing Xu, Wu Jian, Feng Ye, Zhiming Cui · 2008
Traditional level-set methods use gray level images as its main study objects. The basic idea is coupling image data and speed of the curve deformation, stopping the curve evolution on the edge of target, which requires a speed function to control the evolvement. However, traditional methods only set gray-scale gradient information as the stopping power of contour lines to define the velocity function, which will be distort applied to color images. This paper re-designed a speed function based on color gradient function to color images to replace the traditional method of gray-scale gradient to stop the movement of the contour line, and combining regional color statistic characteristics, making the approach is robust against the noise caused by the color gradual change and weakening of the target edge. Experiment shows that accuracy and robust increase significantly in the improved algorithm in color image processing, superior to the original algorithm.