A Study Of Image Features And Target Detection Technique For Holographic Subsurface Imaging Radar

Since holographic subsurface imaging radar(HSIR) has high azimuth resolution, it plays a more and more important role in small shallow hidden target imaging detection. There are more and more demands for such high-resolution imaging probes in the military and civilian domain, such as airport security inspection, medical imaging and archaeological survey.Holographic subsurface imaging radar(HSIR) generally acquires the reflected signals under the conditions of two-dimensional scanning which includes synthetic aperture way and real array aperture way. The acquired amplitude and phase information of the data was used for the imaging processing, and rebuild the target image in two-dimensional azimuth projections or in three dimensions. However, the medium between the radar and the target is too complicated to interpret radar images well, which usually has unknown electromagnetic parameters and complex structure. Therefore, it is essential to carry out the study of image feature and target detection technique for holographic subsurface imaging radar.The study of background and target features is the premise to achieve target detection. When study the background features, due to the unknown density and electromagnetic characteristics of the medium, it’s difficult to obtain an ideal analytic expression. It may be got it through a greater degree of approximation, but this method is more difficult to detect the correct target from the image. From an amount of background images, the specific features of medium can be summarized based on the statistical methods, which can be used for target detection. This thesis presents the background image distribution model based target detection algorithm firstly, and then in order to improve the detection rate and reduce the false alarm rate, the model evaluation factors of the background image were regarded as features to get better results. In order to reduce the dimension of feature set and improve processing speed, the Fisher Discriminant Ratio(FDR) and the Sequential Floating Forward Selection(SFFS) are used to filter features.When study the targets features, this thesis employs a simple threshold-based method to segment image. The texture and geometry features of typical circular target in the homogeneous medium were extracted. From the theoretical analysis of the material, size and location of the target relative to the radar target imaging results for round effects and verified by experiments. As HSIR common goals concern, accounted for a large part of the circular target, and this thesis presents a HSIR image circle target detection algorithm, which is based on the round target image in HSIR peak diameter and circular symmetry three characteristics presented, the algorithm has higher measured data detection rate and low false alarm rate. In order to improve the image SNR, a new algorithm for suppressing medium interference fringe has been proposed.