Research On The Key Technologies Of Image Segmentation Based On Fractal

With the development of information technology, the ability of obtaining information increases quickly. The quantity of various kinds of information becomes more and more large. How to analysis the information quickly, and obtain the interested information is one of the aims to process the information. Image is important information that used widely in many fields. Image segmentation, as one of the important image analysis and extracting technologies, is to extract the contents being interested in, and it is the basic technology of image processing.This thesis deals with image feature vector based on fractal theory combined with image gray level feature.The main results and contributions in this thesis are as follows.1. The fundamental theory and approach of image segmentation was studied based on the fractal theory. The ability of the lacunarity and multi-fractal distinguished different images with similar fractal dimension was verified.2. A novel image feature vector for image segmentation was proposed. Eight fractal features were extracted such as fractal dimension, lacunarity, multi-fractal features from original image and fractal dimensions from transformed image. The 9th feature of image was gray level of some images. The feature vector derived from the features constructed new feature space. The data in feature space been smoothed and filtered were used in image segmentation.3. The program based on K-mean clustering approach were designed and implemented. The proposed method was tested on three types of images (texture image, SAR image and aerial photograph). The results indicate that the image segmentation method based on fractal and gray level features can reduce the impact of noise in image and promote the performance of image segmentation.The conclusion through the different results of the tests is that: the image segmentation method based on fractal is good to texture images, and the image segmentation method proposed by this thesis that based on fractal features combined with gray level feature is good to SAR image and aerial photograph.