| The image is the important carrier which is used to gain information and knowledge from the objective world for human. Digital image processing is always the important content of the computer information processing. By the research, we discover that the image edge contains lots of feature information of the image, and the image edge character is widely used in higher technology and research field, such as image segmentation, image classification, computer vision, pattern recognition and so on. The technology of edge character information extraction is the key technology of digital image processing; the effect of extraction directly relates to the quality of computer vision identification system treatments. Therefore, the image edge's extraction plays an important role in the image processing and it is also the hot points in this field. In recent years, the domestic and foreign experts or scholars amalgamate different new technology and new ways to make it become a very active area.At first, this paper introduced research meaning, situation and the development trend of edge extraction, and then analyses the noise filter methods, conventional differential methods for image enhancement and the respective advantages and disadvantages of differential operator, such as Roberts, Sobel and so on. For the conventional arithmetic of curve fitting image edges, in this paper the curve fitting algorithm will be improved by using striction factor for reducing over-fitting and the image will be enhanced in the fuzzy feature plane, meanwhile, the image will be scanned with block form. Therefore, the image edge feature is highlighted, while its universality is weaken. So, the edge feature will be well positioned and extracted.Finally, experiments show that this algorithm can lead to a higher segmentation quality and the better image edge feature can be gained by the algorithm of this paper. Especially, the edge's veracity and continuity will be well guaranteed. |
PDF Full Text Request