Computer Vision Based Image Detection For LCD Screen Of Remote Controller

With the development of industrial technology, the demanding of automationbecomes more and more higher in production testing, but the LCD screen displaydefects detection of air-conditioner remote controller is still based on manualinspection, works have to spend more intensity labor on this testing, meanwhile, falseand missed detection may occur because of man-made causes. Industrial inspectionbased on computer vision technology which is non-contact, high-precision andefficiency has been increasingly deepened to the detection of industrial products. Inthis paper, we make use of computer vision technology in the LCD screen imagedetection. First, we acquire the image of LCD screen by USB industrial digital camera,and process the image to obtain the numbers, characters and symbol’s edge feature,then, discriminate these feature of the LCD screen image which achieves a good result.The core content of this paper is as follows:(1)Image filtering and sharpening. After acquiring an image with high resolutionby industrial digital camera, we use different algorithm to process number, Chinesecharacter and symbols region for the post-processing requirements. The area ofnumerical we use image filtering to facilitate the numerical recognition processing andimage sharpening to process Chinese characters and symbols area in order to getclearly and continuous edge.(2) Image edge detection. By using different image preprocessing for the LCDscreen image of different regions, edge-detection is a very critical step, because theaccuracy of edge position directly affects the latter part of detection result. In thispaper, we analyze the effect of classic edge detection operator, to get the best effect ofedge detection, we use different edge detection algorithms for different regions of theimage. In digital area we use SUSAN edge detection operator, Chinese characters areais Shen-Castan edge detection algorithm and symbols area is Canny edge detectionalgorithm which have achieved good edge detection effect.(3) Detection software design. Because this paper involved many kinds ofalgorithms, we use modular-design to design and debug the software, then, using theLabVIEW8.6graphical programming language and NI Vision8.6to design detection software which completes camera calibration, image acquisition and detection effectdebugging. At last, we analyze the main factors that affect the detection accuracy andspeed, radial distortion correction of the lens to reduce the measurement error.