Research On Solar Cell Detection Method Based On Multi-source Image

The development and utilization of renewable energy has attracted worldwide attention.As an important direction of renewable energy development,solar energy has been developing well in recent years.Solar cell is the basic unit of solar power generation module,and its quality directly affects the efficiency of power generation.Therefore,it is necessary to strictly detect the quality of solar cell.Existing solar cell detection schemes generally fail to take both surface defects and hidden defects into consideration.In order to ensure the accuracy of detection,it is necessary to use an expensive high-resolution camera to complete image acquisition.Aiming at the problems of incomplete detection schemes and low fault tolerance,a solar cell detection system based on multi-source image is designed.The main research contents of the system are as follows,Firstly,a solar cell surface defect detection algorithm based on visual saliency map and guided filter enhancement is proposed.The algorithm introduces human visual attention mechanism,integrates wavelet transform and frequency-tuning algorithm to extract saliency map of defective target,and further optimizes the global saliency map of the defect target by using guided filter enhancement and morphological closed operation.The global saliency map obtained by this method can not only highlight the information of defect target,but also filter out the redundant background to achieve the goal of defect detection.The main types of surface defects detected include scratches,corners and stains.Then,an algorithm based on visual saliency and NSCT is proposed to detect hidden defects in solar cell.We use a random center-surrounded saliency detective algorithm to extract the local saliency map of the random region of infrared image.Later,the non-subsampled Contourlet transform is performed on the saliency-detected infrared image and the original visible source image respectively,and the different scales and directions are obtained.And then,the high and low frequency subband coefficients are fused by the saliency map guidance method and the absolute value maximization method in the low frequency part and the high frequency part respectively.At length,the reconstructed fused image can be obtained by the inverse nonsubsampled Contourlet transform.Finally,hardware configuration and selection of the system and software design scheme of the detection system are further designed.Image acquisition module of the surface defect detection is composed of low-angle dual LED light source and cameras.The Raspberry Pi connected USB camera is used as the acquisition unit.The image transmission module is based on the TCP network transmission protocol.The hidden defect detection part uses an ordinary camera and a low-resolution thermal imager to complete image acquisition,and we design an APP that integrates the functions of collecting and transmitting images.In addition,this paper also designed a PC image processing software,using MATLAB/GUIDE to complete the development of the solar cell detection system interface.The experimental results show that this system can realize the detection of solar cellaccording to the expected target,and has better detection effect.This system has goodperformance in terms of functional integrity,reliability and cost feasibility,which has a certainengineering value.