Pose detection of partly covered target in micro-vision system

Abstract

Listen

Translate article

In this paper, we focus on the need for location and pose detection of partly covered target. In micro-vision system, the accuracy of location and pose detection directly decides the effect of latter micro-operation. Former methods can't locate the target which was partly covered. According to the current problem in micro-vision system, this paper proposes a method of target location and pose detection which based on Hough transform and template matching. When the aspect ratio and area of target is certain, by detecting the shape feature and using template matching, we detect the location of target which was partly covered. Firstly, use Canny algorithm and morphology method to obtain the edge of objects. Then, use the improved Hough transform algorithm to extract line segments features and locate the candidate target region according to template matching. Finally, recognize and locate the target by detecting whether the two segments are connected to be a signal region. After recognizing target, the slant angle of the target is detected based on Hough transform. This method can exactly locate the target and get the angle. The experimental results show the proposed method is better than the former ones under complex background and partly covered targets.