The Fault Point Of Electrical Equipment Detection Method Based On Multispectral Image Fusion

With the development of the social and economy, the work and life of people is becoming more and more high demand for electricity. How in does not affect the quality of lives and work efficiency of the people, safe and efficient and stable supply of electricity, it is a huge challenge to power workers in China. State overhaul is produced in this background, the state overhaul mainly found equipment safe hidden trouble by charged detection methods, and maintain the equipment of unsafe outage maintenance treatment alone. The fault detection method based on the electrical equipment multispectral image fusion was become an important part of automatic detection system in the smart grid. Therefore, it has important research significance and application value which design an efficient electrical equipment state detection algorithm.In this paper, for problem in the fault detection method based on the electrical equipment multispectral image fusion, we propose a fault detection method based on the hypercomplex Log-Gabor filter and hypercomplex visual saliency. Our works are as followAccording to the problem which the general image filter can only filtering the gray image, this paper proposes a design method of hypercomplex Log-Gabor filter. This filter can filter the color image as a whole, to extract the image feature.According to the problem which different spectral image segmentation. This paper proposes different segmentation method for different spectral image. The segmentation of visible image is based on the hypercomplex Log-Gabor filter method, first of all the quaternion of visible image convolve with hypercomplex Log-Gabor filter, and then extract the local quaternion phase of the convolution results, finally we use the threshold segmentation and edge detection for the local quaternion phase; The segmentation of infrared image is based on visual saliency. First we use the hypercomplex Log-Gabor filter to filter image, and then extract the quaternion Fourier transform phase spectrum of the filtered image, finally we use the Mean Shift and adaptive threshold segmentation method to extract saliency region. Experimental results show the proposed method is superior to others method and can accurately extract the fault point.According to the imaging characteristics of the different spectral image. This paper proposes a novel wavelet transform image fusion method based on the edge and the maximum coefficient. The low frequency component adopt based on low frequency domain edge selection strategy. The high frequency component adopt based on the maximum absolute value coefficient. Experimental show that:the proposed algorithm can accurately detect the fault region in the natural image.In conclusion, this paper proposes a fault point of the electrical equipment detection method based on multispectral image fusion consist of detection and extraction and fusion. We propose different segmentation method for different spectral image and extract the fault point by fuse the segmentation image based on wavelet transform.