Research On Ship Target Detection Technology Based On Polarimetric Image Sequence

In modern war,anti-ship missile and shipborne electronic warfare defense system are contradiction,which promotes the continuous development of jamming and anti-jamming methods.For example,the corner reflector can form echoes that are very similar to the ship target in time,frequency and space through the combined structure and array arrangement,which makes it difficult to detect the ship target concealed under the interference of the corner reflector by the method of High Resolution Range Profile(HRRP).Therefore,by analyzing the characteristic differences between ship target and corner reflector in polarization domain in actual combat countermeasure environment,this paper carries out the research on the method of polarimetric SAR image registration,moving ship detection and anti-corner reflector interference technology based on polarimetric image sequence.The main research contents of this paper include the following three aspects:1.As the basis of image processing of polarimetric SAR,this paper first introduces the characterization and preprocessing methods of polarimetric SAR image,focusing on the geometric correction methods of polarimetric SAR image based on multi-look processing,and compares the performance of several commonly used filtering methods for polarimetric SAR image in filtering speckle noise.Then,local invariance feature extraction algorithm is a common algorithm for image registration.The core of this algorithm is to detect grayscale changes in images by constructing gradient operator.To make full use of the polarization change information,this paper studies several commonly used polarization changes detection algorithms.This part of the research provides a theoretical basis for the later construction of the polarization gradient operator model.In addition,several typical local invariance feature extraction algorithms are studied,and their limitations and shortcomings are analyzed.2.To solve the problems of inadequate utilization of polarization information and the strong influence of residual speckle in the image registration process by the existing polarization SAR local invariance feature extraction algorithm,a polarimetric SAR-SIFT algorithm based on the polarization gradient operator is presented.The core of polarimetric SAR-SIFT algorithm is to construct a polarization gradient operator that is not easily affected by speckle based on polarization change detection.The polarization gradient operator improves the construction of scale space and the calculation of neighborhood gradient direction histogram in traditional SAR-SIFT algorithm,The polarimetric SAR-SIFT algorithm can make full use of the polarization change information to generate more different feature description vectors,and can extract the structures that are not significant in the single polarization image but significant in the full polarization image,so as to improve the accuracy of image registration.Through the measured data,it is verified that the polarimetric SAR-SIFT algorithm proposed in this paper can get good image registration results even when processing images with large viewing angle difference.3.To solve the problem that the typical passive interference styles,such as ship target and floating corner reflector array,are difficult to recognize in the real combat environment,this paper presents a moving ship detection method based on polarimetric image sequence.The algorithm first completes registration of continuous multi-frame images based on the polarimetric SAR-SIFT algorithm proposed in this paper,then obtains the velocity difference of moving ships in continuous multi-frame images relative to the static background by extracting the change information from the images,and finally detects moving ships by speed-position density clustering.Due to the comprehensive utilization of the changing information of moving target in multi-frame images,this algorithm can effectively filter out the influence of floating corner reflector array on ship target detection and can resist the same kind of corner reflector interference,which is of great military significance in the combat environment.The effectiveness of the algorithm is verified by combining the measured and simulated data.Finally,the content of this paper is summarized,and the limitations of this method and the work that needs further study in the future are discussed.