An Improved Active Contours Model Based on Morphology for Image Segmentation

An improved active contour model is proposed for image segmentation in this paper based on the theory of mathematical morphology. With a structure element shifting on contour points, the pull and restraint force to contour points can be calculated through morphology operations. The contour points move toward and locate on the boundary of the detected objects with the control forces. The experimental results show that this improved active contours model can overcome some shortages of highly sensitivity to initial curve position, hardly convergence to concave shapes. More importantly, Image segmentation based on the improved model can extract more accuracy contour shape and texture feature in application of surface defects detection, and significantly improve the accuracy of defects classification.