A novel sub-pixel edge detection for micro-parts manipulation

This paper presents a novel sub-pixel edge detection and center localization algorithm to improve the resolution and accuracy for micro-assembly systems. There are three steps to realize the algorithm. First the invariant moment is used to describe the feature of micro-parts and the edge of the micro-parts is extracted at pixel level by using Canny operator. Secondly, the image information of gradient direction from the edge neighborhood is employed to locate the edge of the micro-parts at sub-pixel level. Last, apply least squares ellipse to get the position of the parts centers. Experiment results show that the presented algorithm could achieve higher location accuracy and consume less time.