Sonar Image Segmentation And Recognition Based On Fractal Theory

Along with increasing exploitation of marine resources, the intellectualized requirements of sonar equipments is getting stricter, more and more researches are put in the field of acoustic vision information processing. Among all the researches, fractal theory breaks a new path for sonar image processing. The results of studying show that fractal feature, of which measure fixity is applicable to sonar image segmentation and recognition technologies has been found in natural texture image. The works of this paper are given below:1,General methods of estimating fractal dimension in image processing are analyzed, also applicable range and computation of different methods are compared.2,Sonar image segmentation methods based on different fractal features are deeply studied, then an improved sonar image segmentation method DBC combined with Peleg-εblanket method is given. By analyzing the methods, advantages and disadvantages are concluded.3,Sonar image recognition algorithms based on different fractal features are deeply studied. First, fractal dimension is used as texture feature. Then, BP network is used as classifier to recognize the images. Finally, by comparing with the traditional methods, the recognition method in this paper is validated.