A novel color machine vision system development for color classification and AR film thicknesss measurment of silicon crystalline solar cells

• Main Authors: WANG, Hao-chien, 王浩鑑
• Other Authors: 顏旭男
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/04753233971247950993

碩士 === 聖約翰科技大學 === 電子工程系碩士班 === 98 === For increasing marketing competence, silicon crystalline solar cell manufacturers are forced to adopt computer vision inspection techniques in production lines to perform product classification and statistical process control (SPC). The product classification is based on the overall photoelectric conversion efficiency of the solar cell itself. Two factors directly influence the overall photoelectric conversion efficiency of the solar cell, i.e., the composed materials and the anti-reflection (AR) film coating on substrate. Since the thickness variation of anti-reflection thin film will induce the color change on the surface of solar cell, this study develops a cost-effective color computer vision inspection system to perform fast color classification and AR film thickness measurement of silicon crystalline solar cells. The proposed system first uses color charge coupled device (CCD) to capture the red-green-blue (RGB) color image of inspected silicon crystalline solar cell, and transforms it to hue-saturation-lightness (HSL) image format. And then the area and boundary of different hue-value images are calculated and sorted with the image thresholding and label operation. Besides, with the corresponding measurement procedure on specified hue-value regions of using a precise height measurement instrument, such as the wavelength scanning profiler, the regression equation between the hue value and AR film thickness is obtained, and then implemented into the color vision system to perform large area scanning AR film height measurement of silicon crystalline solar cells. Compared to the optical ellipsometry, the measurement speed of the proposed system is fast. It can take only 0.1 seconds or so to finish the height measurement of a 12.5cm×12.5cm solar-cell image, and the measurement accuracy can reach 30Å.