Soda Casket Inspection System by Computer Vision

Abstract

Referred to the problem a manufacturer always faces that bottles in a casket are missing, caused by the bottles leaning down. Checking the parceling by human causes high risk from the explosion of the lean-down bottles. So this article presents how computer vision is applied to the inspection system. The system used a low-price webcam and image processing method to identify the defection. The process begins with adjusting HSV up before working on edge detection to reduce density of data. Then cross correlation as template matching to define the bottles and check the number of soda bottles. This system can identify 84% of completely loaded casket and 100% of the missing loaded casket.