Institute of Electrical and Electronics Engineers (IEEE)
The performance of image segmentation depends on the output quality of the edge detection process. Typical edge detecting method is based on detecting pixels in an image with high gradient values, and then applies a global threshold value to extract the edge points of the image. By these methods, some detected edge points may not belong to the edge and some thin edge points in dark regions of the image are being eliminated. These eliminated edges may be with important features of the image. This paper proposes a new mathematical morphological edge-detecting algorithm based on the morphological residue transformation derived from dilation operation to detect and preserve the thin edges. Moreover, this work adopts five bipolar oriented edge masks to prune the miss detected edge points. The experimental results show that the proposed algorithm is successfully to preserve the thin edges in the dark regions.
Signal Processing and Information Technology, 2004. Proceedings of the Fourth IEEE International Symposium on, pp.310-313