Area Segmentation of Images Using Watershed and Anisotropic Gaussian Kernels

Abstract

Image segmentation as one of the primary steps in image analysis for object identification has received much attention lately, the main aim being to recognize homogeneous regions of an image which belong to different objects. A noise-robust edge detector SCHEME based on anisotropic Gaussian Kernels (ANGKs) is proposed in this paper by combining the region-merging algorithm to obtain higher quality segmentation results, which consist of three basic blocks. The first two blocks give rise to an initial partition of an image, while the last block attains the final segmentation by iteratively merging similar fragmented regions. A series of experiments is used to evaluate the performance of the method. The experimental results show that the proposed method has good noise robustness and localization accuracy, and can solve the problem of over-segmentation effectively, the simulation result supports our proposed main ideas.