Template Matching of Round Features of Aluminum Alloy Castings Based on Hu Constant Moment

Abstract

Automobile structural parts, that is automobile frame, are used as load-bearing equipment with advanced mechanical properties so that it need to be strictly inspected. There are geometric characteristics, uneven distribution of positions, and large volume for automobile structural castings of aluminum alloy. Thus, it is necessary that the feature matching deals with the automobile structural casting by detection of visual image. However, the traditional visual image has exposed many essential problems, such as lower degree and lower field-of-view stitching accuracy. According to the above phenomenon, a template detection method is proposed based on Hu moment, combined with the target of round feature under the front sub-frame of a certain model of Zotye Automobile. The picture is converted into a binary image with the connected domain label. The feature values are obtained by the template image and the regional features is calculated. Secondly, the Euclidean distance is applied to judge the similarity before finishing the template matching. Finally, the matching accuracy can reach 100%, but also running time of the algorithm is less than 0.08s for the round holes of the front sub-frame under experimental data.