Method for automatically splicing scrap images through computer

Abstract

The invention provides a fragment matching method based on contour curvature. The method comprises steps as follows: firstly, fragments are pre-processed, a curvature string of a contour curve of each fragment is acquired, then each closed contour curve is segmented into curve edges by the aid of contour corners, regular fragment boundaries are rejected, and finally, longest matched curve sections among contours are found out with an LCS (longest common subsequence) algorithm and spliced. According to the method, on one hand, the matched curve section search range is defined by the aid of the contour corners of the fragments, so that the time for searching the matched contours is shortened, and the time complexity of the algorithm is greatly reduced; on the other hand, the regular fragment boundaries with constant curvature are rejected, so that mismatching is reduced, and the accuracy of the algorithm is improved.

Description

technical field [0001] The invention relates to a splicing technology of two-dimensional image fragments of irregular objects. Background technique [0002] Many fields will encounter the problem of splicing fragments. For example, in archaeological research, due to the large number of fragments and the unknown original shape of the broken cultural relics, it will be very difficult or even impossible to restore the original appearance manually. Therefore, the research on the automatic splicing of fragments is of great significance. [0003] In the past two decades, scholars at home and abroad have conducted a series of researches on the problem of splicing fragments, and put forward some effective methods. Generally, there are two main steps to realize the automatic mosaic of fragments: the first step is to find all the adjacent fragment pairs (also called matching pairs) in the original object by using information such as the color, texture or outline of the fragments; the...

AI Technical Summary

Problems solved by technology

However, this algorithm has high requirements for contour extraction and sampling, and is difficult to apply in practice

As another example, the fragment matching algorithm proposed by Wolfson et al. using the idea of ​​substring patchwork is simple and stable in calculation, but this method can easily lead to mismatching because it cannot provide accurate information about fragment matching; Leitao et al. search for fragments by continuously refining the sampling scale The local matching algorithm is effective and stable, but the time complexity is too large

Images