A new method for image matching

To achieve accurate measurements of large scale machine parts with machine vision techniques, a mosaic method for image sequences is selected. And the key technique of mosaic is image matching, which mainly affects measure speed and precision. This paper aims at improving measure speed. The proposed method bases on a straight profile. It mainly consists of two stages: One is adjusting image sequences to the same direction of coordinate axis to eliminate rotating deviation. It is realized by searching the straight profile with chain code. The other is sequences matching to eliminate shifting deviation. The new algorithm focuses on the eliminating shifting deviation. It makes good use of differences of nearest neighbor gray values. Two templates are taken separately from two neighbor sequences. Those templates complete matching by the minimum value of the sum of adjacent-pixel-difference. In experiments, the new method is contrasted with the classical algorithm of the normalized cross correlation method. Experimental result demonstrates that the proposed algorithm is more effective than the NCC for image mosaic.