| Synthetic Aperture Radar (SAR) is a high-resolution radar which can work uninterrupted at any time and under any weather conditions so that it has been widely used in various fields of national life and national defense. However due to its special coherent imaging mechanism, there are a large amount of speckle noises in SAR images, which bring a serious impact on the subsequent target detection and identification. Therefore, how to remove the speckle existing only in SAR images effectively is an indispensable content in the field of SAR image processing. In view of the fact that our country is rich in marine resources, and has a vast sea area, it can protect the marine rights and interests of China better to strengthen the monitoring of the target in China’s sea area. With the more and more extensive and important application of synthetic aperture radar in the ocean monitoring, the research on ship detection in SAR images is a very interesting subject.Firstly, this dissertation simply expounds the coherent imaging theory of SAR images, and analyzes the formation principle of coherent speckle noise only in SAR images further. Aiming at the two research contents in this paper, it introduces the traditional SAR image de-noising and ship detection algorithms respectively. Then two algorithms are proposed in this paper based on above basic theory. The main work is as follows:One algorithm is an iterative filtering method on the de-noising of synthetic aperture radar image based on block matching. Firstly, it makes use of the similarity among the image blocks to conduct block matching and to construct a three-dimensional matrix of the similar image block. And then, it performs iteration filtering to remove speckle noise in transform domain among similar image blocks, so that we can obtain the basic estimate de-noised image after reconstruction by weighted average. Finally, it performs block matching on the basic estimate de-noised image, and filters in the similar image blocks and among the similar image blocks to get the final de-noised image by3-D experience filter. It compares the proposed algorithm with other algorithms on the real SAR images through experiment. Experiments show that the proposed method can effectively suppress speckle noise of SAR image, while preserving better the edge and texture information.The other one is a ship detection algorithm based on non-local means of two-parameters CFAR. Firstly, the image will be changed into windows, and these detection sliding windows are divided into the target window, the protection window, and the background window. The next, calculating the Euclidean distance between the detection window and each image window, it gains a number of similar tiles and uses these similar tiles to conduct the operation of non-local means on the pixels in background window. The whole image information is used to estimate the two-parameters of local sea clutter background and set the false alarm rate. Finally, calculate the detection threshold of the, conduct the binaryzation individually on the pixels to be detected in the target window, and gain the final detection results of the ship. It compares the proposed algorithm with other algorithms on the real SAR images through experiment. Experiments show that the proposed method gets a good effect not only on the missed rate but also on the false alarm rate. |
PDF Full Text Request