Research On Key Technologies Of In-situ Image Measurement System For Light Guide Plate Micro-structured Punch Machine Tool

In recent years,the rapid development of liquid crystal display technology has led to the development of backlight module in downstream industry.As the key component of backlight module,the quality of liquid crystal display is determined by its manufacturing technology.In order to improve the processing technology level of the light guide plate micro-structure lattice,our research group developed a micro-structure punching machine tool,which needs the assistance of measuring device to process the micro-structure point array of a given size.In order to solve the problem that traditional measuring instruments and machine tool workbenches are not on the same physical platform and that the measurement requires repeated loading and unloading of workpieces,which leads to low processing accuracy,this paper proposes a scheme of integrating in-situ measurement system on micro-structured punch machine tools based on machine vision technology.The scheme is studied from two aspects: hardware and software algorithms.The following are the main contents of this paper:Firstly,the overall design of in-situ measurement system is carried out according to the structural characteristics of light guide plate micro-structured punch machine tool,and the selection of hardware and software,system detection algorithm and other aspects are analyzed.In view of the characteristics of visual measurement,this paper studies the pre-processing algorithm of micro-structure image,including image filtering,binarization and mathematical morphology processing.The noise distribution of micro-structure image is studied by histogram,and bilateral filtering algorithm with better protection of edge details is selected.Otsu algorithm for threshold selection of micro-structure image is improved,which reduces the scope of threshold search and improves the search speed.Mathematical morphology technology is used to remove the fine marks and speckles that interfere with target detection.Pretreatment brings convenience for subsequent edge detection.For the pre-processed micro-structure images,this paper compares and analyses the classical pixel-level edge detection operators from two aspects of positioning accuracy and running speed,and proposes a sub-pixel detection algorithm based on Roberts-Zernike moments.Compared with the traditional Zernike,the detection speed is greatly improved on the premise of ensuring the accuracy.In view of the small field of vision of sub-millimeter field measurement system and the inapplicability of traditional camera calibration technology,this paper proposes a two-step calibration method based on checkerboard calibration board,which separates lens distortion coefficient and CCD pixel equivalent.The distortion coefficient of lens can be deduced by taking a chessboard image,which is easy to operate and has high accuracy of calibration results.Finally,the stability of the detection system is analyzed through the processing experiment of the micro-structure point array.By comparing with the measurement results of the white-light interferometer,the conclusion that the maximum error of the system in this paper is less than 1.4?m is drawn.The reliability of the system as an assistant processing method of light guide plate micro-structured punch machine tool is verified from two aspects of stability and accuracy.