| Summary: | 碩士 === 國立臺灣科技大學 === 自動化及控制研究所 === 99 === This study focuses on developing an automatic defect detection technology of a solar cell. A line scan CCD with high speed, high resolution, and vast inspecting scope is utilized to detect the surface defects by integrating image processing methods. This study employs the improved SUSAN corner detection algorithm to detect the defects such as crack, pinhole etc. In order to properly set the thresholds of SUSAN corner detection algorithm, the correlation between different geometry and different threshold is investigated to obtain the relation of different corner angles and geometry. For the threshold setting of gray scale differences, the improved SUSAN corner detection algorithm is use to analyze image gray value and image contrast. Thus, the gray scale difference threshold is adaptively selected according to different image contrast. This method is evaluated with an octagonal poly-silicon solar cell images grabbed by a line scan CCD or a scanner. The experimental items include scratch, fracture, pinhole, and crack. From the experimental results for parameter settings of different corner detection, it shows that the best results can be obtained by using geometry threshold of 18.
|
|---|
