Enhancement of image edge definition by modified Quadtree sobel operator

This paper presents an image segmentation method based on modified Quadtree sobel operator which was applied to enhance the precision of defining the real edges of the digital images. After removes noise of the images via modified median filtering, and pre-segments the images according to their characteristics, quadtree operator merges different zones without noise by multiple-zone target segmentation algorithm. The segmented images has clear structure and reserves their characteristic information. Results show that the method can removes noise effectively and solves out the problem of inaccurate segmentation due to regional image overlapping.