The Research Of Imaging And Target Detection And Identification For Sas

With the further exploration and utilization of the ocean, the technique of underwater acoustic imaging and the sonar image detection and recognition have become an important subject in the field of digital image processing. To solve the problem of the integration of high-definition imaging sonar, this paper makes a deep study on the imaging, detection and identification of underwater objects of SAS (Synthetic Aperture Sonar). The main contents include:(1) The paper gives a review of the internal and external present situation of sonar technique and introduces the composition of high-definition sonar system and main technical means at present. And the technical characteristics and properties of SAS have also been introduced. Besides, the paper analyzes the technical difficulties in the current sonar image processing.(2) LFM (Linear Frequency Modulation) Signal in the SAS system is introduced, as well as the basic theory of its form and processing method. The paper also illustrates the pulse compression of LFM signal and the theory of the high definition of SAS, which lays a solid foundation for the subsequent imaging process.(3) As for the concrete problem of SAS imaging, the classical RD (Range Doppler) Algorithm is adopted. In order to simulate the imaging of real underwater objects, the 3D modeling of typical objects such as mines, torpedoes, and submarines is necessary to realize the fast area-target imaging, so that the simulation of complex objects could be accomplished efficiently.(4) In the light of the 2D Azimuth-Range images of underwater objects derived according to RD Algorithm, the improved CFAR (Constant False Alarm Rate) Algorithm is presented to perform the object detection. The histogram of the real image is used to grossly replace that of the background clutter to be fitted to the distribution of common clutters. Then the parameters of the distribution are calculated and used in the CFAR detection to obtain the detection threshold, which is critical to the get the final detection result.(5) On account of the detection result of SAS objects based on the improved CFAR Algorithm, the Chain-code Disperse Curvature Algorithm is put forward to identify the objects. The definition of chain-code in the Pattern Recognition being standardized, with the discretization of curvature in the calculus, a pattern that suits the description of digital images is constructed. Then the two are combined, and different objects are classified according to the method of collection mappings, so that the final identification is achieved.