The Quality Detection Methods Of Film Panel Based On Machine Vision

With the improvement of modern industrial levels, increasing production speedand production quality requirements but lower detection capabilities have become animportant factor restricting the development of the production of the film surface panel.Now the production the production of the film surface panel is also commonly used bybusinesses traditional manual detection method-under strong light, with peopleobserving with the naked eye to detect, which on the one hand increases the company’sinvestment and the impact of artificial vision health of monitoring workers,anotherbecause of the subjective aspects of human factors testing process by the impression,as a results,in uneven test results difficult to obtain satisfactory results. In recentdecades, due to the continuous development of computer hardware and software, thetheory of machine vision matures, domestic and foreign scientific and technical staffwill use their respective technologies to industrial inspection up.In this paper, we use the technology based on machine vision to detect the versionof the film surface, therefore designed a film surface panel detection system. Thelighting system consists of an imaging system, an image acquisition system (within therange of image transmission and communication related issues do not belong to thisstudy, it is not in this article focuses on), image processing software system. Throughthe production of the film surface panel image acquisition and testing, to determine thefilm surface panel printing quality is satisfactory, to achieve better results.This image processing algorithms ideas for the design of the film surface panelprinting quality testing software as follows: gathering the gray images of film surfacepanels which are in high quality when it is in off line time,pick the best one as templateimage;panel will gather to face off version of the real-time image based on thedirection of the gradient created by offline vectors derived form this shape matchingpose corresponding image and the offset template image, and then use this offset to the real-time image conversion panel, and then using the subtraction method and thetransformed detectde image for comparison using threshold segmentation methodwhich will show the original flaws that exist in the real-time image, and then we usethe flaws to determine whether the detected panel threshold for eligibility bycomparing the size of the area and set the calculated apparent defect images.