Color Difference Detection And Color Classification Of Photovoltaic Cells Based On Machine Vision

In the photovoltaic industry,color difference refers to the uneven distribution of color on the surface of the same photovoltaic cell.Color difference detection means the detection of the uniformity of color distribution on the surface of the same photovoltaic cell.The color difference detection algorithm directly affects the quality control of the next production link component product.In this paper,the characteristics of the surface color of the photovoltaic cells are analyzed.Color difference detection algorithm based on chromatic aberration histogram is used for feature extraction and chromatic aberration detection.The best color difference characteristic parameters are obtained,and the color difference detection of photovoltaic cells is realized,and the problem of color difference detection is solved.Different colors of photovoltaic cells are assembled into components,which will reduce the beauty of components,affect the power generation efficiency of components,and reduce the service life of components.So the color classification of photovoltaic cells with different colors have great value and significance.Aiming at the existing problems of color classification and recognition of photovoltaic cells,including the problem of small color difference and weak discrimination ability among different kinds of photovoltaic cells,two algorithms are deeply analyzed and put forward.One is to improve the photovoltaic cell classification algorithm based on the improved Gauss hybrid model.The other is a photovoltaic cell classification algorithm based on polychromatic space features and DAG-SVMs.The specific research contents are as follows:(1)Aiming at the color difference problem of photovoltaic cell,a color difference detection method based on chromatic aberration histogram is proposed.By extracting the color difference histogram of the L*a*b* color space,the chromatic contrast ratio of the photovoltaic cells is improved.The extracted features are quantized and distance measured,and the optimal parameters are obtained.The problem of color difference detection for photovoltaic cells is effectively solved.Some experiment are carried out on the TJYL-PV-CD.The algorithm achieves good detection results on the image data set.The effectiveness of the proposed algorithm for chromatic aberration detection of photovoltaiccells is verified.(2)Aiming at the color classification problem of photovoltaic cells,a color classification algorithm for photovoltaic cells based on improved Gauss mixture model is proposed.Using the RGB color model which is close to the human visual system,the statistical analysis of the color features on the RGB color space is carried out,and the Gaussian mixture model is improved by remove the isolated points.In order to verify the performance of the method,the experiment was carried out on TJYL-PV-AB.The results show that the method is satisfying.(3)Aiming at the color classification problem of photovoltaic cells,a color classification method of photovoltaic cells based on polychromatic space feature fusion and DAG-SVMs is proposed.The feature filtering method of multi-color space visualization is proposed,which effectively solves the problem of contrasting with different color spaces,and the feature selection is quick and accurate.The first application of DAG-SVMs theory in the color classification direction of photovoltaic cell visual inspection.In order to verify the effectiveness of the algorithm,the experiment was carried out on TJYL-PV-AB.The results show that the proposed method has good classification results.