Study On The Restoring Technology Of Irregular Planar Fragments Based On Shape Feature

Irregular fragments restoring as an important research is of widespread practical use and attracts attention of scholars at home and abroad. On the basis of the study of Image Processing, Pattern Recognition, Differential Geometry and shape matching technology and previous work, this paper presents a new method based on shape feature for irregular planar fragments restoring, uses outlines to represent fragments shape and then converts fragments restoring into curve matching and curve merging. The research work can be summarized in the following respects:Firstly, fragment outlines are extracted with a tracking algorithm and a smoothing method is applied on the contour curve with a Gaussian function as its variance varies from a small to a large value, the result is the multi-scale space image of fragment. According to the basic theory on curve and the shape representation criteria, curvature descriptor is introduced to describe outlines.Secondly, considering that there are a large collection of randomly-mixed fragments to be restored in practice, this paper develops a special multi-scale matching algorithm to search matching fragment pairs. Compared with single-scale matching method, it allows us to reduce the total computation largely. Then, based on the matching pair information, a valid ergodic algorithm for the sequence of fragments merging is presented. In the algorithm, the locating of fragments in the merging list is determined by the current matching-fragment and another fragment, which ensures the process of fragments merging. According to the merging list and the coordinate transformation formula for points, virtural demonstrate on fragments merging is realized. Besides, repairing methods are presented for sorted lost fragments.Lastly, a prototype system for fragments restoring is designed by analyzing the modular structure, and an example is given to check the validity of the restoring algorithm elementarily.