Studies On The Defecting Inspecting Technology Of Solar Panel Based On Machine Vision

With the rapid development of society,the demand for energy is increasing.The traditional fossil energy is non-renewable energy,and there will be some pollution to the environment.As a clean and renewable energy source,solar energy has gradually entered people's field of vision and has gradually been widely used.Solar panel as a mold for the production of solar cells,the quality of solar panel will directly affect the quality of solar cells.At present,most solar panel manufacturers still use manual observation methods to detect defects and measure the size of grid lines on solar panels.With the continuous transformation and upgrading of manufacturing companies,the requirements for product quality,measurement accuracy,and detection accuracy are getting higher and higher.Traditional methods of human eye observation are increasingly difficult to meet the requirements for the detection accuracy of solar panels.In recent years,with the continuous development of machine vision technology,more and more companies use machine vision technology to detect product defects,mainly including size measurement,defect detection,and positioning.The use of machine vision technology for product defect detection has the characteristics of high speed,high accuracy and so on.Based on the above characteristics,this paper makes use of the machine vision method to research the defect detection of solar panels.The specific research contents include: design a hardware platform suitable for the image acquisition of solar panel;Using digital image processing technology to carry out grid line positioning and size measurement;Using a support vector machine to detect defects in solar panel images;In order to display the entire solar panel image,stitching technique was used in the processing of solar panel images.The hardware system is mainly used for taking pictures of solar panel images,mainly including the design of mobile platform,selection of industrial cameras,lenses,and light sources.The software system mainly includes the grid line size measurement and defect detection.Firstly,the solar panel image is denoised,and the median filter is used to filter the solar panel images.In the digital image processing technology,using image threshold segmentation,image centroid extraction,image projection and straight line fitting method to achieve accurate positioning of the grid line.On the basis of the positioning of the grid line,we can realize the grid line size measurement of the solar panel.Solar panel defect detection is mainly implemented by the image classification method of support vector machines.The support vector machine classifier is designed by the creation of training sample images,the extraction of HOG features and the selection of support vector machine parameters.Based on the positioning of the grid lines,a welldesigned classifier is used to detect the defects of the solar panel.In order to display a whole solar screen images,the screen images need to be stitched.We have improved the technology of the feature points detection area and matching of solar panel images,and the improved algorithm is applied to the SIFT stitching algorithm to achieve solar panel images stitching.