Research On The Fusion Detection Method Of Infrared And Visible Light Targets

Information fusion is different from traditional data integration,which acquires stable and reliable comprehensive information through reasonable coordination of multi-sensor.The target detection technology of low SNR ratio has been widely used in military warning,tracking and monitoring systems in recent years.In order to achieve the purpose of target fusion detection under the condition of weak visible light,we combine the infrared targets with visible light backgrounds not only to highlight the infrared targets but also to retain details of visible light images through the fusion detection of infrared and visible light targets.This paper which focused on the fusion detection of infrared and visible light targets was divided into three main parts:fusion detection,fusion recognition and fusion strategy.Firstly,based on the theory of infrared background prediction,a fusion detection scheme was proposed and an improved Tri operator was defined to replace traditional detection operator for target edge extraction.Sets of experiments have been designed to guarantee the detection accuracy and identification clarity.At the same time,observers can also get the background information.Then in view of limitations of traditional method of recognition,a target recognition scheme based on hybrid classification feature was proposed.CSF,Hu and HOG feature were respectively selected from visual,edge and gray scale categories as the hybrid classification feature after establishing the sample library.Multi-scene pedestrian recognition with different features(groups)and fusion recognition combined with visible light images experiments were conducted based on SVM classifier training.The result proved that we can identify the targets accurately and obtain ample environmental information as well with this scheme.Afterwards,a fusion strategy based on target segmentation was proposed in order to solve the defect of traditional pixel-by-pixel structure.According to FD principle,the relative sharpness of gray-scale fused image was improved.Aimed at gray scale similarity issue between infrared target and visible light background,two solutions of color fusion FD structure and improved FD structure were put forward.Finally,based on MFC in VS2010 platform,a fusion detection software of infrared and visible light targets is designed.