Design And Implementation Of Ship Target Detection Algorithm Based On SAR Image

Synthetic Aperture Radar(SAR)can detect the target area for a long time and a long distance in different climatic conditions,and it has been widely used in marine monitoring,disaster monitoring,surveying and mapping,military applications and other fields.At present,the ship target detection technology based on SAR images has become an important research content in ocean monitoring,which has extremely important research value and significance.The high sea conditions and sea conditions in the complex sea battlefield environment lead to too many interference factors in SAR images,which affect the performance of ship target detection.Traditional detection algorithms have some problems,such as poor accuracy of target feature extraction and low efficiency of algorithm execution.Aiming at the above problems,this paper studies the ship target detection technology based on SAR images,and the main work and innovations are summarized as follows:1.Research on SAR image preprocessing methods,including speckle suppression and sealand segmentation.Firstly,the generation mechanism and noise model of SAR image speckle noise are studied,and several classical spatial domain filtering algorithms based on statistical model and new non-local mean filtering algorithms are introduced.The filtering performance of each algorithm is verified and analyzed by actual image simulation.Then,the algorithm of maximum inter-class variance,which is commonly used in land-sea segmentation,is further studied.In this paper,the algorithm of two-dimensional maximum inter-class variance is applied to automatic land-sea segmentation without prior knowledge.Simulation results show that this algorithm can obtain a purer sea surface area.The SAR image preprocessing method used in this paper can effectively improve the image quality and reduce the clutter interference.2.Research on ship target detection algorithm in SAR image.Firstly,the classical constant false alarm rate(CFAR)algorithm,several common clutter statistical models,corresponding parameters and threshold estimation methods are introduced.Then,aiming at the problems of low detection rate and poor real-time performance of traditional CFAR detection algorithm in complex environment,a two-level CFAR detection algorithm based on superpixels is studied,which is divided into global detection and local detection.In the local detection,the traditional clutter window has a fixed size,which can’t adaptively select the clutter area,resulting in a large number of selected clutter samples,which affects the efficiency of the algorithm.In this paper,the superpixel neighborhood window is used to select the superpixel blocks in the superpixel neighborhood of the target to be detected,and it is classified according to the similarity characteristics,so as to select a proper amount of clutter samples and effectively improve the selection efficiency of clutter samples.The simulation results show that the improved two-level CFAR detection algorithm based on superpixels selects homogeneous superpixels in the neighborhood,which makes the parameter estimation of clutter model more accurate and effectively improves the detection rate and running efficiency of the algorithm.3.Multi-core parallel implementation of SAR image speckle suppression,sea-land segmentation and target detection module algorithm based on multi-core DSP platform is studied.Firstly,aiming at the problems of long processing time and low utilization rate of system resources in single-core system,this paper designs a multi-core and multi-task parallel processing SAR image target detection scheme based on data flow model.Then,the task of SAR image ship target detection algorithm is assigned to run in different cores.Finally,the system resources occupied by each algorithm are planned and allocated on the chip architecture of TMS320C6678,and the correctness and feasibility of the detection scheme are verified on the development board of TMS320C678 by combining the technologies of inter-core synchronization mechanism and memory optimization.Compared with single-core processing,the multi-core processing scheme in this paper can effectively improve the program running efficiency.