应用独立成分分析的太阳能电池图像去噪

发布时间：2018-11-10 11:09

【摘要】：研究用于太阳能电池缺陷检测的电致发光图像噪声去除的问题。太阳能电池电致发光图像是近红外图像,一般来说采集到的图像具有噪声严重、亮度不均匀等特点,使用快速独立成分分析(Fast ICA)将图像分离成两个独立成分,一个是噪声,一个是所需的分离结果。由于分离结果中仍存在椒盐噪声,对分离结果再进行均值滤波,最后得到理想的太阳能电池图像。实验结果表明,此方法对太阳能电池电致发光图像的去噪效果显著,为进一步的图像处理做好了准备。
[Abstract]:The problem of noise removal from electroluminescent images used for defect detection of solar cells is studied. The electroluminescent image of solar cell is near infrared image. Generally speaking, the collected image has the characteristics of serious noise and uneven brightness. The image is separated into two independent components by fast independent component analysis (Fast ICA), one is noise, and the other is noise. One is the desired separation result. Because there is still salt and pepper noise in the separation result, the mean filter is applied to the separation result, and finally the ideal solar cell image is obtained. The experimental results show that the proposed method is effective in de-noising the electroluminescent images of solar cells, and is ready for further image processing.
【作者单位】：北京工商大学计算机与信息工程学院;
【基金】：北京市教委科研计划面上项目(KM201210011003)
【分类号】：TP391.41;TM914.4

【参考文献】