Synthetic Aperture Radar;Segmentation;Level Set;Active Contour;Fourth Order Cumulant;Genetic Algorithm (GA)

Gholamreza Akbarizadeh

《International Journal of Remote Sensing Applications》 2卷4期 (2012/12) Pp. 7-19

並列摘要

Image segmentation is an important step in texture recognition of Satellite Synthetic Aperture Radar (SAR) images. In this paper, an efficient method for texture recognition of SAR images is proposed based on extraordinary result of wavelet transform on texture feature extraction and the benefits of genetic algorithm (GA) as a classifier. First, texture feature is extracted by wavelet transform. Second, a feature vector composed of wavelet energy feature, kurtosis value of wavelet energy feature and gray values of eight-neighborhood of SAR image is formed. Then, segmentation of different textures is applied by using feature vector and level set function. Finally, the values of feature vector are used for training the classifier. Two different classes of datasets are used to show the experimental results. One is a simulated SAR image and the other is a real satellite SAR image. The results of experiments show the powerful ability of this algorithm to recognize different textures in Satellite SAR images.

並列關鍵字

Synthetic Aperture Radar ； Segmentation ； Level Set ； Active Contour ； Fourth Order Cumulant ； Genetic Algorithm (GA)