A lens collar auto-inspection system

Surface quality of lens collar arouses customers´ highly concern. Manual inspection has been a way for quality assurance, but it might cause employees injuries and strenuosity. So it is worthy to introduce a nondestructive inspection system. The proposed system is composed of a CCD, a white LED coaxial light and a motor. The region of interesting (ROI) of lens collar is extracted firstly where is filled with statistical texture. The discrete cosine transformation (DCT) image restoration sub-function reflects textures into high energy components which are then be oppressed by filters. We finally attempt to highlight the defects by defect detection sub-function because of gray levels of textures within a range of limit while defects not.