Abstract: In view of the low accuracy and slowness of traditional SIFT (Scale-Invarivant Feature Transform)algorithm of tool image matching in complex environment, a SIFT image matching algorithm based on edge information is proposed. The Canny operator is improved by adaptive smoothing and improved Otsu algorithm.
The edge information of the tool image is obtained. The edge is expanded and closed by morphological operation, and the edge of the edge is removed. The SIFT algorithm is used to extract the stable feature points to achieve the matching of the tool image. The experimental results show that the improved Canny operator can eliminate the influence of the pseudo edge on the edge detection and have strong adaptability. For the tool image with scale change, scaling, rotation and partial occlusion, the improved algorithm is superior to traditional SIFT algorithm in terms of matching accuracy and matching time. Which is better than the traditional SIFT algorithm.

Key words: scale-invarivant feature transform (SIFT) algorithm;image matching, Canny operator, morphology, Otsu algorithm