Automated cutting process may produce superficial defects such as scratches and depressions on lens collars which are caused by cutter offset or scrap accumulation. Additionally, electroplating defects, such as dots and uneven electroplating occurs if the surface is rough or contaminated with foreign matters. As the inclined camber of lens collar contributes to external appearances of a single-lens camera, customers extremely concern about its surface quality. Relying on human inspection to ensure the quality is time-consuming, labor-intensive and produces occupational injuries. Therefore, the implementation of auto-inspection can help overcome these problems. The inspection system is composed of charge coupled device (CCD), coaxial light and motor. After an image is taken, it goes through a segmentation sub-function to obtain a region of interesting (ROI). Since the texture of inclined camber of a lens collar is statistically distributed, an image restoration sub-function based on discrete cosine transformation (DCT) is used to map the texture onto high-energy components on a spectrum. They are then compressed by a notch-rejecting filter. In contrast with the defects, the grey value of texture is limited within a certain range. Statistical process control binarisation, curve fitting and binary large object (blob) analysis are used to highlight defects.