Research On Infrared Image Segmentation Algorithm And Its Application In The Fault Diagnosis Of Electrical Equipment

Image segmentation is to divide the image into various non-overlapping regions with different characteristics and extract the interested parts.It is an important part of digital image processing.It helps feature extraction,image measurement and image display.It is the basis of image analysis and one of the most difficult problems in computer vision.Since the 1970 s,many experts and scholars have attached great importance to the research of image segmentation algorithm,and achieved gratifying results in the past decades.It has been widely used in various fields,such as biomedical image processing,pattern recognition,document image processing,production process control,remote sensing and biomedical image analysis,and agriculture.The results of image segmentation directly affect the performance of image analysis vision system.At present,a variety of segmentation algorithms have been proposed,but there are still some problems such as the lack of edge direction information extraction,edge breakage,discontinuity and non-smoothness.At the same time,when the edge of the target is blurred and the background information of the image is complex,it is difficult to extract the target and the extraction effect is not ideal.Therefore,the research of image segmentation has always been a hot and key point in the research of computer image technology.Usually,the fault diagnosis of electrical equipment is to measure the operating parameters of the power system by means of the instrument,and to draw a conclusion through calculation and analysis.This method is not only inefficient in operation,unable to achieve the purpose of real-time fault detection,but also complex in operation,consuming manpower,material and financial resources.Electrical equipment faults often occur with thermal effects,so infrared thermal images can be used to analyze electrical equipment fault images,so as to quickly troubleshoot electrical faults.For the basic segmentation algorithm of infrared thermal image analysis,the main work of this paper is as follows.(1)In view of the fact that the traditional edge detection operators often fail to detect the multi-gradient information of the edge in the infrared power equipment images with complex background,which contain more scenery for long-distance shooting,and the phenomenon of edge breaking,non-smoothness and discontinuity will appear,a scalable multi-gradient image segmentation method based on edge detection is proposed.An improved algorithm is proposed based on the traditional first order edge detection operator: Sobel operator.It increases the number of the direction template to eight,and preserves the multi-direction gradient information of the image.At the same time,the scaling factor is used to scale the gradient information,which makes the edge of the extraction smoother and more continuous,and it improves the accuracy of edge detection.The effect of edge extraction is experimented and analyzed.(2)In view of the problem of low accuracy in fault area extraction of infrared electrical equipment with complex background and blurred target edge,an image segmentation method based on maximum entropy detector is proposed.It uses the region growing method to locate the target according to the characteristics of the image we process,thus eliminating the background interference.The images of the target and the background are generated by using the maximum entropy principle,which are used for training the detector.Finally,the maximum entropy segmentation image is selected by using the detector of the target and the background.Finally,the target which is experimented and analyzed is extracted.(3)In view of the low contrast and low signal-to-noise ratio of infrared images,and the problem that the extraction of multi-target infrared electrical equipment faults is often ineffective,a new image segmentation method based on k-means clustering and region growing is proposed.And it is used to deal with the fault image of electrical equipment.The method grays the fault image of infrared equipment firstly.Then because the k-means algorithm is very sensitive to the noise,the filters are used for denoising.We use the k-means clustering algorithm to get the seed points of region growing method.Then carrying out region growth by utilizing the seed points,and finally obtaining the target extraction image.And the results of the experiment are analyzed.