Sub-pixel Detection Algorithm Based On Silicon Gate Line Detection System

Solar become more and more important in the development and use of new energy sources.The core part of the solar is photovoltaic cell.Silicon wafers for an important part of solar cells,if a defect occurs in the solar cell production process,it will affect the photoelectric conversion efficiency of the optical fiber and increasing production costs.Based on this thesis scenarios,the thesis proposes sub-pixel detection algorithm based on silicon gate line detection system,it has important theoretical significance and application value.Combined with current research related to the application of technology,the paper analyzes the detection system of silicon gate line position in the solar cell industry,resolve the problem that current machine vision algorithms to detect the presence of the detection accuracy can only achieve pixel-level problem.The thesis has improved the algorithm for multiple related program modules:In terms of the image gray distribution is uneven,the paper proposes the adaptive histogram equalization to strengthen the details of image,image noise has been effectively suppressed and the clarity of the image has been improved.In terms of the surface characteristics of solar wafers,the thesis proposes mean iterative and sobel operator method to improve the positioning accuracy of the edge detection.In terms of image threshold selection in the coarse matching process,the thesis proposes searching image septal point mode in order to find rough match point.In the precise matching process,the thesis regards the resulting crude matching correlation point as the center point,then use quadratic surface fitting method to find the optimal matching point to improve the overall processing speed.In terms of sub-pixel edge detection problem,the paper compares and analysis many sub-pixel edge detection algorithm,and selects the best method and improves it.After the test verification,the detection accuracy is 0.2 pixels.The main contributions of this dissertation are summarized as follows:●A adaptive histogram equalization method is improved,in order to enhance the details of the image,the background noise is effectively suppressed and image layering gets improved.● A threshold segmentation algorithm is improved,the iteration has been used in binarization,so that the running time of the module is reduced to 1/3 of the original.● SSDA template matching algorithm is improved,the operation time is reduced and the matching accuracy is improved.