Investigation On Appearance Inspection Of Solar Cells Based On Machine Vision

With the increasing demand of energy, solar energy is being widely utilized since it is clean, green and renewable. Generally, solar cells require a strict quality test to check the problems of physical damage, surface scratch, sharpness of printed pattern and so on. In the traditional industrial model, these detections are achieved manually, resulting in a low efficiency and wrong detections due to the visual fatigue and subjective assume of employees. Fortunately, with the rapid development of computer technic and digital image processing, it is feasible to apply the machine vision to detect the solar cell defections.This paper firstly analyzes the disadvantages of the image segmentation methods which are simply based on a single threshold. Due to ununiform brightness and low image contrast, a single threshold migght cause mistakes in the result of image segmentation. In order to solve this problem, a more effective image segmentation method is proposed. The new method combines the advantages of the local thresholding methods and the region growing methods. Experimental results show that the new method has better robustness and can accurately extract hidden appearance defects.Since a solar cell module is usually welded together with a series of solar cells, the precision of camera calibration is especially important for the vision detection system. Unfortunately, the wide-angle lens used for image capture suffers from barrel distortion. In this paper, the linear interpolation method is used to eliminate the lens distortion, and the least-square method is used to fit a straight line in order to reduce the position error of coordinate.Finally, we build a vision detection system of solar cells on the windows operating system. Experimental results show that the proposed defection detect method and the coordinate calibration method are not only feasible but also precise.