In this paper, we develop a real time vision inspection system for the product in automated manufacturing site using the computer vision technique. To be a successful system, we must solve two problems: (1) A good quality of image for moving object. (2) Real time vision inspection algorithms. To solve those problems, we study the optical reflection characteristics of PET material for minimizing the local unbalanced strong reflection, and select the best lighting device for a good quality of image. We propose two pre-processing algorithms for minimizing the searching area and processing time. We also propose the inspection algorithms to inspect the seal surface of PET bottles. The system consists of position detecting sensors for the moving object, computer, CCD camera, strobo lighting device, and air gun for removing the bottles with defect. We make the user interface program for the changing parameters and sensitivity, and the inspection program in digital signal processing board.