Research On Railway Foreign Object Intrusion Detection Method Based On Multi-source Image Fusion

With the rapid development of the high-speed railway in China,the detection technology of railway foreign object invasion is very important to ensure the safety of high speed rail operation.At present,there is no effective way to solve the problem of false detection and missed detection caused by the day shadow interference and the quality of night images.In this paper,a detection method of railway foreign object invasion based on multi-source image fusion is proposed to compensate for the deficiency of single visible image detection under the shadow and night conditions by using infrared and visible image fusion.First of all,the overall scheme of the railway foreign object invasion algorithm based on multi-source image fusion is proposed.Then,the characteristics of infrared and visible images and the basic theory and technology of the related algorithms are introduced.On this basis,this paper focuses on the registration and fusion algorithm of the heterogenous image and the multi source image fusion detection algorithm of moving targets.Finally,a series of daytime and nighttime experiments verify the performance of the algorithm.According to the problem of high mismatching rate of feature points in heterogeneous images,an improved registration algorithm based on SURF feature points is presented.In order to get rich and detailed color information and highlight the target temperature characteristics,a fusion algorithm of Contourlet transform based on T V model and local energy is proposed.Aiming at the problem of false detection and missed detection caused by the infrared halo interference and light mutation at night,a multi source information fusion detection method for moving target based on foreground-background discrimination factor is proposed to realize the accurate detection of the moving target while the enhancing the visualization and comprehensibility of the image.Compared with the commonly used algorithms of image fusion,the fusion image obtained by the fusion algorithm proposed in this paper has a significant improvement in the quality of the source image.Compared with other fusion algorithms,Compared with other fusion algorithms,the standard deviation increased by at least 12.6%,the information entropy increased by at least 10.5%and the cross entropy decreased by at least 65%.The detection results of the moving target fusion detection algorithm in the day and night are better than the moving target detection results of single source video.It solves the problem of inaccurate detection of moving objects and false detection in the condition of daytime infrared halo interference and night light mutation.It is of great significance for the accurate detection of railway foreign object intrusion and the improvement of railway safety.