Study Of Moving Target Detection Method For Through-wall-radar Based On Visual Attention Mechanism

Through-wall-radar system,which has the capability of detection,imaging,tacking of concealed objects behind the nonmetal obstacles such as walls,brushwood,relic etc,by utilizing the penetration ability of low-frequency-band electromagnetic wave,are playing an important role in a variety of military and civilian applications recently.However,on the one hand,confined by the real practice,because of the existence of obstacles,the electromagnetic wave of target has to experience dispersion,multi-reflection and refraction when it travels inside of the obstacles.On the other hand,considering the penetration attenuation,the electromagnetic wave echo has so much wall clutter and environment clutter,which is often non-homogenous and non-uniform,that make target information invisible.Additionally,electromagnetic wave traveling inside of the obstacles facing multipath and shadow effect,which results in that traditional through-wall-radar detection processing methods,are difficult to achieve ideal effect.In order to improve the quality of target detection,my dissertation based on the analysis of the traditional radar detection technology analyses the reasons of false target.And a new target detection method based on the visual attention mechanism theory is proposed for through-wall-radar system with the step frequency arrays and multiple-input-multiple-output arrays in the dissertation.Finally real data collected by the experimental have been used to validate the efficiency of the proposed method.The major works of this dissertation are as follows:First of all,this paper analyzes traditional through-wall-radar detection techniques,including the change detection technique,the back projection in time domain algorithm and constant false alarm rate algorithm.To analyze the impact of the ghost target for the detection results,we establish point target,extended target and moving target models to analyze the situation how ghost target formSecondly,on the basis of biological structure of the vision system and neurology,we describe the information processing of human's vision system and analysis some related characters in visual attention mechanism.Besides,according to the image features from through-wall-radar system with MIMO arrays,we introduce a new algorithm framework based on top to bottom mechanism and the Itti visual attention mechanism model.Based on the framework mentioned above,we build up a data-driven model from bottom to top,which transforms the completed images into two sub-characters,intension and Doppler.After that,we combined a new synthetic map based on the main-appurtenant technique.Besides,we search all map and extract the focus of attention using the shift-prohibit mechanism.Thirdly,considering the low SCR in the detection of the through-wall-radar system,we introduce detecting progress using top to bottom system.At first,we propose the pre-attention processing to control the focus,which can focus on the area in which it interests and build up a center of temperate memory and knowledge database.Besides,we use the prior information from moving model of human and analysis the different conditions of single target and multi-targets.Then,we realize the real-time detecting and tracking in different conditions,which improves the quality of the images effectively.Finally,the dissertation makes a summary and gives the further research.