An automatic seeded region growing algorithmbased on contour processing for grayscale images

Abstract

In this paper, an automatic seeded region growing algorithm based on contour processing for grayscale images is proposed. In which, the grayscale image is first read. After that, the image is subjected to contour processing. The contour image is then segmented using simple SRG by assigning labels to each pixel in a specific segment. The segmented image is then compared to original input image to find the suitable seed in the segment depending on the average brightness of pixels forming the segment. Then these seeds are used again to generate segments using SRG algorithm. The proposed algorithm is shown that it is more accurate in finding number of segments than traditional segmentation algorithms.