Research On Crack Detection Algorithm For Polycrystalline Solar Cells

With the continuous exploitation and consumption of traditional energy,the energy crisis has become a serious problem faced by all countries in the world.Therefore,many countries pay more and more attention to the development and utilization of renewable resources.Solar energy,as a clean and environmentally friendly renewable resource,has been favored by countries all over the world in recent years.At present,the development and utilization of solar energy is mainly to convert light energy into electric energy through solar cells.Therefore,the quality of solar cells will directly determine the efficiency of photoelectric conversion.Therefore in the process of development and use of solar cell,the solar cells defect detection plays a very important role,and in all the defects,due to the cracks are very small and hidden,detection is the most difficult,the traditional method of crack detection is mainly by the human eye detection,costly manpower.This paper mainly realizes crack detection of solar cells by image processing technology,including the following aspects:(1)In order to solve the problem of perspective distortion of solar cells in the process of shooting,we design an automatic vertex determination method based on geometry to replace manual labeling of the vertices of distorted areas in the image,so as to ensure the automation and batch quantification of crack detection of solar cells.(2)Before crack detection,the pre-processed image must be edge extracted,but the complex background texture of solar cells causes great interference to edge extraction.In this paper,a multi-scale edge extraction algorithm combining texture suppression and Laplace pyramid decomposition is proposed,which can effectively suppress the complex texture background in solar cell images and locate the edge region.(3)In the process of crack detection,a fuzzy c-means clustering crack detection algorithm based on edge features is proposed.A new crack feature descriptor is defined by combining the haar-like descriptor of crack morphology features and the wavelet energy difference method of neighborhood texture features.The feature vectors of edge points of flawless solar cell images are clustered by dichotomy fuzzy c-means clustering algorithm.The clustering results are compared with the feature vectors of the edge points of the images to be detected,and the counterexample is identified as cracks.This method solves the problem of not directly extracting fracture features,and can effectively suppress noise and retain fracture information as much as possible.(4)Because of the influence of background image,the obtained cracks will be discontinuous.In this paper,an improved directional zone growth method is proposed to optimize the crack detection results,which can make the crack detection results more complete.