Research On The Technologies Of Computer-aided Image Fragments Reassembly

Reassembling unknowing broken objects with the assistance of computer from a large collection of irregular image fragments arises in several fields such as archaeology, criminal detection, paleontology, conservation ( mural painting), and so on. It was impossible if there were no computers to assist in reassembling these tens of thousands of randomly shaped and featureless image fragments.2D image fragment matching is deeply investigated in this thesis. The methods of pretreatment and reassembly of image fragments are introduced. And we research more on image fragments matching method which is the core of image fragments reassembly. A method for 2D contours matching based on angle-sequence is presented in this thesis, which can be used to obtain the best pairwise matching of image fragments. Multiscale technique is used with our method to improve computational speed. Compared with curvature-sequence, it not only resolves the problem of sharp corners in the process of image fragments merging, but also improves computational speed. Finally, the image fragments are reassembled with program of Matlab.