Research On Moving Target Detection Algorithm Based On Image Fusion

Moving target detection is undoubtedly the foundation and priority among priorities of today's intelligent video monitoring system,as well as the premise of subsequent target identification,tracking and behavior analysis.However,moving target detection is easily affected by complex scenes such as illumination,haze,rain and snow.In order to improve the robustness and applicability of the system,visible light and infrared imaging characteristics are combined to obtain better moving target detection effect by combining their advantages.This paper focuses on the application of computer vision technology in intelligent video monitoring,and studies the visible light and infrared image enhancement,registration,fusion and moving target detection algorithm,so as to cope with the complex scenes in the monitoring and realize the all-weather and all-time work of the monitoring system.First of all,this paper describes the commonly used image enhancement algorithms,analyzes the visible image and infrared image imaging characteristics,and according to their respective characteristics to enhance the image quality and highlight the region of interest,so as to lay a foundation for the subsequent image processing.Then the image registration method based on feature is introduced,the feature points are extracted and matched by visual significance and SIFT method,the registration of heterologous images is completed by affine transformation and bilinear interpolation,and the invariance and robustness of this method are verified in many cases.Then this paper analyzes the common visible and infrared image fusion methods,puts forward the color image fusion method based on spatial frequency and fuzzy transformation,and evaluates and compares the advantages of this method from the subjective and objective aspects,which lays a good foundation for the detection of moving objects and better deals with complex scenes.At last,the common algorithms for moving target detection are compared,and the proposed method is combined with the improved inter-frame difference method and the background subtraction method of W4.The proposed method is verified and compared in different scenarios,which shows the advancement of the proposed method in both qualitative and quantitative aspects.