Sar Image Of The Ship Target Extraction And Fractal In The Sar Image Classification

Target detection and classification in SAR images is one of the importantaspects of the remote sensing application. In this paper, we study the detection ofship targets in SAR images, and the classification of SAR images with fractalmethod.Ship Detection in SAR images has a broad prospect of application. Firstly, weanalyze the features of ship targets in SAR images, and compare some methods withdifferent filters and different thresholds. Moreover, we discuss the shape features andtheir quality calculation, which could be used for the target c1assification andrecognition later. Finally, an algorithm with a series of steps is presented for the shiprecognition.In image analysis, Texture information is very important, and fractal method iseffective in texture analysis. We study the application with fractal method in targetdetection and classification. In addition, we also analyze the infection of multi-lookprocessing and de-speckle filtering.We focus on our research in two sides: sea-land segmentation and man-madeobject detection. Compared with other related methods, fractal method can do well inthe sea-land segmeotation. With the fractal theory man-made objects usually havevery low fractal dimension while that of natural background is usually high. Basedon this difference, we could distinguish the man-made objects from naturalbackground. In the end of this paper we give some experiment results that showfractal method is effective and adaptive to the detection of different kind of man-made objects.