Distribution Of Target Detection Method Based On The Physiological Visual System Of Marine Sar Image

Remote sense on ocean using SAR has been widely attended at present. There are large numbers of distributed targets in the surface of ocean, such as the wake of ship etc. It is an important content to study the character of distributed targets and the method to detect the features The major aim of this paper is to put forward an algorithm to detect the character of distributed targets in SAR images in order to realize automatic detection of distributed targets and classify the character primarily.Firstly, this paper introduces the basic theory and the primary character of SAR image. And then we discuss the imaging theory and the imaging character of the distributed targets in SAR ocean images. On the basis of above discussion, some existing major methods to detect the wake of ship are listed, and every method is compared with others in the aspect of limitation and the processed effect.The effect of the detection of the wake of ship in SAR images primarily lies on the different of radar backscatter variations between the region covering the distributed targets and the background sea surface. Usually, because of the less contrast between the distributed targets and the background sea surface and the influences coming from the status of the ocean, the kind of the ship, moving parameters and the observation conditions of SAR, the targets is easily covered over by the noise. It is more difficult than detection ship (ship appeared in SAR image as strong targets). At present, the detection of ship wake is focus on the linear character such as V-shaped Kelvin arms and a turbulent wake. However, the situation of the ocean and the physics state of ship is so complex that the reality ship wake appeared in SAR Ocean is a complex mixture of speckle and zonal character. Up to now, few productions has published in this area and there are many things to do.This paper puts forward a method to detect non-linear character. This method uses Static Boundary Contour System based on biological vision to process image in order to pick-up the direction information of wirelike character by the way of competition and cooperation. And then process the image by morphological filterbased on mathematical morphology. After discussion the theoretical feasibility, the effectiveness of this algorithm is demonstrated by the test results of simulating wake images and actual SAR images of ship wakes. The results indicate this algorithm can detect non-linear distributed targets in SAR image effectively. The collateral feature of this algorithm, which is easy to realize by hardware, show the potential of practicality.In the course of graduate study, I also accomplish the network image database for the need of laboratory. And it has been well used.