Research On Key Technology Of Infrared Detection Of Dim And Small Target Under Complex Background Conditions

The detection of infrared dim and small targets under complex background has great theoretical significance and practical application value,especially in the military field,which is often used in infrared reconnaissance,guidance and early warning.Due to the characteristics of small area,low brightness and weak contrast with the surrounding background,in the original infrared image,the highlight background,background edge and noise under the complex background are easy to submerge the real infrared small target,which leads to the problems of low detection rate and high false alarm rate in the current detection algorithm for the infrared small target under the complex background.At present,based on human visual system(HVS)and local contrast,the principle of infrared dim and small target detection algorithm is easy to realize and fit the original infrared image features,which has attracted a lot of attention from the scholars.Inspired by the idea that HVS is more sensitive to the contrast information in the field of view than the brightness information,this kind of algorithm improves the detection rate and reduces the false alarm rate by improving the contrast between the local small target and the background in the infrared image,and also has the multi-scale detection ability.Therefore,in order to solve the problem of infrared dim small target detection in complex background,this dissertation will focus on the research of a new single-frame and multi-frame infrared dim small target detection algorithm based on human visual system(HVS)and local contrast mechanism.Although more researchers have been engaged in the research of infrared dim and small target detection algorithm based on human vision system,but on the whole,many problems have not been well solved,mainly including the following three problems: firstly,under the complex background conditions,the existing HVS single frame algorithm in the definition of local contrast often does not enhance the target and suppress the background clutter at the same time;secondly,the existing HVS single frame algorithm has less research on the auxiliary weighting idea under the complex background condition,which fails to make full use of the local background complexity weighting information to further suppress the background clutter,and the detection rate of small and weak targets with very low signal to clutter ratio is often low;thirdly,the research on the multi-frame space-time algorithm has just started in recent years,and the existing multi-frame algorithm under the complex background condition,the spatial domain and time domain are usually calculated separately,the disadvantage is that the 3D spatial-temporal domain is not considered as a whole.The fusion of the spatial domain and time domain will increase the false alarm,and the detection performance of any part will affect the overall detection performance.In this dissertation,we first analyze the characteristics of complex background and infrared dim small target,and analyze and summarize the typical infrared dim small target detection algorithms in recent years from two aspects of single frame and multi frame.Then,aiming at the shortcomings of infrared dim small target detection algorithms,we propose new infrared dim small target detection algorithms based on single frame and multi frame under complex background small target detection algorithm: firstly,an infrared small target detection based on facet-kernel filtering local contrast measure(FFLCM)is proposed,this method uses the facet-kernel model to fit the gray-scale surface of the center block in the infrared small target image to enhance the target.At the same time,combining with the similarity background suppression algorithm,it starts from the two aspects of enhancing the target and suppressing the background to realize the detection of the infrared small and dim target.Secondly,a new single frame detection algorithm under complex background conditions,namely,multi-scale homogeneous-weighted local contrast measure(HWLCM)is proposed.In this algorithm,a new local contrast calculation method is proposed,and on this basis,a homogeneity weighted calculation factor is proposed to further suppress complex background clutter and enhancement goals.Thirdly,a new multi frame detection algorithm based on 3D space-time domain spatial-temporal local difference measure(STLDM)is proposed.This method proposes a new 3D local image block structure by using the local contrast change feature of the target center and the local contrast change feature of the surrounding area of the target in the multi frame 3D space-time domain The two features of contrast variation are used to detect the local contrast of the target in eight directions directly in three-dimensional space-time domain.The above two algorithms are verified by comparative experiments.Experiments show that our three methods,especially HWLCM and STLDM algorithms,can not only significantly enhance the target,but also effectively suppress the background clutter.They have good detection performance for the infrared dim and small target with very low signal to clutter ratio under complex background.