The research of power equipment's on-line monitoring based on monochrome CCD sensor is achieved to improve the shortcoming of infrared thermal imaging instrument such as its complicated structure and the high price. The image processing module of the system is researched. Based on infrared monitoring theory, the experiment system is constructed and the near-infrared thermal images are collected. The processing of image is accomplished by using independent component analysis and many image fusion methods. A new algorithm combined Newton's iteration method is proposed on the basis of studying FastICA. The simulation shows that using the improved algorithm, the convergence speed can be increased with the correspondent separation performance. A new method of obtaining the order of ICA's separation signal is proposed. This method not only can obtain the order, but also can extract the interested independent component. Now the lowest temperature can be detected by this system is 130℃. To some extent, this system can replace infrared thermal imaging instrument. |