Research On Key Techniques Of Small Target Detection For Airborne Infrared Search And Track System

Airborne infrared search and track system can quickly find and lock enemy targets,effectively enhance the aircraft's ability to perceive the battlefield situation;its high precision of angle and distance measurement can provide accurate information support for weapon strike system;the passive detection principle has strong anti electronic warfare interference and good concealment performance,which can improve the battlefield survival ability of aircraft,and is an important component of modern integrated avionics system.The premise of"strike first"is"find first",in order to find the target as early as possible,detection distance is the key index of airborne IRST.Under the condition of long-distance imaging,the size of the target in the infrared image is small,the signal intensity is weak,which make detection difficult.Due to the complexity of the imaging scene,the target is also vulnerable to noise and clutter interference,which reduces the operational efficiency of airborne IRST;in addition,the battlefield environment is complex and changeable,which puts forward higher requirements for the adaptability of detection.Based on the engineering development of the new generation airborne infrared search and tracking system,this thesis studies the small target detection techniques of airborne IRST,and is committed to improving the detection ability of airborne IRST target detection system for long-distance targets in complex background,and enhancing its scene robustness.The main contents of this thesis include the following aspects:(1)The infrared imaging preprocessing method is studied.Good imaging quality is the basis of high performance infrared small target detection.The factors that affect the quality of infrared imaging are analyzed,and the methods of nonuniformity correction,invalid pixel compensation and random noise suppression are mainly studied.Based on the analysis of typical nonuniformity correction methods and the engineering application of airborne IRST,a nonuniformity correction method of two-point calibration combined with real-time calibration of migration coefficients is proposed,and a complete set of airborne IRST imaging preprocessing technology scheme is formed,which can improve the image quality.(2)The basic theory and method of infrared small target detection are studied.The characteristics of infrared small target image are analyzed.The key characteristics of infrared background and small target(local continuity and non local correlation of background,local saliency and global sparsity of small target)are summarized.The characteristics of infrared small target detection task are defined,and the general framework of infrared small target detection algorithm is described.The small target detection model based on infrared patch is introduced,the low-rank of background tensor and the sparsity of target tensor are clarified,and some related mathematical concepts are introduced,which are the basis of following research.(3)An infrared small target detection method based on Gaussian scale-space enhanced local contrast measure is proposed.Inspired by the characteristics of human visual system,the method combines Gaussian scale-space with local contrast,calculates local contrast on scale image,and obtains gray features of local area of image directly through down sampling,which makes local feature extraction more reasonable and efficient,and has better background suppression and target enhancement effect.At the same time,the method utilizes the extreme attribute of small target,designs a new local contrast calculation form,and matches the size of sliding window with the scale factor,so as to obtain excellent detection performance,high computation efficiency.The method has high engineering application value.(4)This thesis propose an infrared small target detection method based on tensor average rank non-convex surrogate.Based on the infrared patch tensor model,the method starts with the measurement of the low-rank characteristics of background tensor and the utilization of local prior information.The background tensor is constrained by the partial sum minimization of average tensor nuclear norm and the target saliency map obtained by the local contrast method in Gaussian scale space is used as prior information to weight sparse target tensor,which alleviates bias of the background separation caused by nuclear norm,combines the advantages of local prior and non-local prior detection methods,and accelerates the convergence speed of the model.A large number of experiments show that the method has superior performance in typical application scenarios of airborne IRST.(5)An infrared small target detection method based on Laplace function non-convex tensor rank surrogate is proposed.In the method,the Laplace function of singular value is used to measure the low-rank property of background tensor,which can better approximate the₀norm of singular value and achieve better background-target separation effect;a local contrast energy feature is proposed as a priori information fusion into the detection model based on infrared patch tensor;the structural sparse regularization is introduced to further suppress those clutter with sparse characteristics.The method can significantly enhance small targets in complex background,suppress various forms of background clutter interference,and has strong scene adaptability and anti-noise ability.(6)We develope a small target detection system based on multi-core DSP+FPGA.Considering the engineering practice of airborne infrared search and tracking system,the joint detection strategy is proposed,the real-time signal processing platform is designed,the test environment is built,and the system is tested.The technical methods proposed in this thesis have been verified by extensive experimental tests,perform well in comparison with similar methods,and improve the performance of small target detection of airborne IRST under complex background,and some specific problems in engineering application are solved,which provide strong support for the engineering development of new generation airborne infrared search and tracking system.