Research On Algorithm Of Color Difference Detection And Color Classification For Polycrystalline Silicon Cells

Polycrystalline silicon cells are widely used in the solar photovoltaic industry.However,in the actual production of polycrystalline silicon cells,not only unqualified cells with defects such as cracks,chipping,breakage,color differences,etc.,but also qualified cells with different colors,which will seriously affect the efficiency of power generation and the beauty of the components,so it is necessary to test the cells.Surface detection technology based on machine vision is widely used because of its non-contact,fast detection speed,etc.The detection algorithm is the key technology to realize machine vision detection.This paper studies the color difference detection and color classification algorithm of polycrystalline silicon cells.The main research contents are as follows:(1)Image preprocessing.In order to provide high-quality image support for subsequent color difference detection and color classification,firstly,the characteristics of the original cell images are analyzed.Aiming at the tilt problem of the original image,the tilt correction of the cell images is realized by threshold segmentation and edge contour detection.Aiming at the interference of external features such as conveyor belt,the location of the cell is determined by constructing vertex coordinates to achieve efficient extraction of the cell area.(2)Color difference detection based on maximum area contrast.Aiming at the problem of color difference detection of polycrystalline silicon cells,this paper proposes a detection method based on maximum area contrast after analyzing the characteristics of color difference defects of polycrystalline silicon cells.The method separates the color difference suspicious area from the background area of the image by color space conversion,channel separation,threshold segmentation and region contrast.Based on this,the feature extraction and characterization of the cell image is realized.Finally,the classification algorithm is used to identify the color difference defect of the polycrystalline silicon cells.The effectiveness of the maximum region contrast algorithm for color difference defect detection is verified by The experiments of image segmentation and color difference defect recognition.(3)Color classification based on tabu genetic convolution neural network.In order to improve the color classification accuracy of the cells,the tabu genetic algorithm is proposed to adjust the hyperparameters after analyzing the basic principle and architecture of the convolutional neural network model and the influence of hyperparameters on the whole model.The method of obtaining the optimal network model is performed,and the obtained optimal neural network model is used to classify the polycrystalline silicon cells.Based on the open data set and network architecture,a comparison experiment with other hyper-parameter optimization algorithms is carried out to verify the effectiveness of the tabu genetic algorithm for the hyper-parameter adjustment algorithm,and to evaluate the speed and accuracy of color classification using the cell data set.At the end of this paper,based on the above algorithm,the design of color difference detection and color classification system is completed,which can realize the operations of reading the battery image,detecting color difference,classifying color,and statistical detection category.