Research On Image Detection Technology Of Dynamic Small Dim Targets

The detection and tracking of infrared dim target is widely used in the military and civil fields such as precision guided,video surveillance.Infrared early warning system has advantages of invisibility,high maneuverability,no electromagnetic interference,can work in all weather and so on.But under the complex background infrared image gray fluctuation,target pixel accounted for less,low signal-to-noise ratio and lack of shape and structure information,resulting in infrared dim target detection and tracking is much more difficult than the visible light image.For the above problem,according to the process of the infrared dim small target detection and tracking,the image preprocessing,target detection and tracking algorithms of infrared dim small target are analyzed and studied in this paper.And we implement infrared dynamic dim small target detection and tracking algorithm in VS2010 software platform.Firstly,this paper analyzes the characteristics of infrared dim small target image,and established the mathematical model of infrared dim-small target image.Three kinds of noise reduction methods and two kinds of image enhancement methods are compared in this paper.Theoretical research and experimental results show that the performance of the maximum median reduction method and the enhancement method of the gamma transformation is higher than other methods.Secondly,considering that the infrared small dim target detection and tracking system has certain requirements for the real-time and accuracy of the algorithm,this paper effectively combines and improves the sequence image detection algorithm based on the inter-frame difference and the sequence image detection algorithm based on background difference.For the problem of the segmentation threshold of the context and target that changes with the environment,an adaptive threshold is introduced as a basis for background and target segmentation.The experimental results show that the improved infrared dim small target detection algorithm greatly improves the detection rate and reduces the false alarm rate and still satisfying the real time requirement.Finally,in order to solve the problem of tracking failure after the target is obscured by the KCF tracking algorithm,the KCF algorithm and the improved infrared dim small target detection algorithm are effectively combined,and an improved infrared dynamic dim small target detection and tracking algorithm based on kernel correlation filtering is proposed.For the false alarm phenomenon of target detection,the cumulative mean of motion distance is introduced to calculate the tracking results,thus further reducing the false alarm rate.The experimental results show that the improved infrared dynamic dim small target detection and tracking algorithm based on kernel correlation filtering can realize the infrared dim small target tracking in real time,accurate and stable,and can effectively solve the problem of occlusion of the target.