Moment-preserving edge detection and its application to image data compression

Edge detection plays an important role in many applications, such as industrial inspection, pattern recognition, scene analysis, image coding, etc. Since human visual perception is sensitive to edges and contours, we can simplify shape analysis and perform more accurate object recognition if the boundary of an object can be extracted successfully. In addition, image data compression is crucial to reducing the amount of storage and transmission time in the applications of database management and network communication, respectively. Therefore, some effective data reduction techniques that do not destroy object shape boundaries are desired. The goal of this study is to develop a new edgedetection technique with subpixel accuracy that can be used in image data compression. For a survey of edge-detection techniques with "pixel" accuracy, see Ref. 1. The gradient and the Laplacian are two