Optical Lens Edge Detection Method Based Coaxial Alignment Micro Assembly

Abstract

Optical lens assemblies are key to affecting lens accuracy. The coaxial alignment could improve the accuracy of lens assembly effectively in automatic assembly system. But it was difficult to capture the full edge of the lens in one image because of the limit of the CCD camera field of view and the image edge of the lens is not sharp. The purpose of this paper is to propose an edge detection method of optical lens based on Holistically-Nested Edge Detection (HED) for a micro-assembly system with coaxial alignment function. First, the edge of the optical lens is got by HED method, and then the thick edge of the detected optical lens is refined image and the pixel points in the middle position of the thick edge of the recognized image is extracted. Finally, the least square method is used to fit the edge image of the lens. The results of comparative experimental shows that this method which can extract edge of the optical lens image automatically is more robust and more accurate than the previous edge recognition methods.