Single channel SAR image segmentation using gamma distribution hipothesis test

Segmentation is a low-level operation extremely important for the extraction of information from digital images. A commonly adopted approach for the development of segmentation algorithms is that based on statistical modeling of data. For SAR images, the precise knowledge of the statistical properties and the adoption of an appropriate modeling are considered essential to obtain reliable results. Depending on the type of data and the degree of homogeneity of the imaged target, different statistical distributions can be used to obtain an appropriate modeling. Within this context, this paper aims to present the results obtained in the segmentation of a single channel SAR data in the intensity format using two different algorithms. The first, called SegSAR [1] uses the Gaussian and Gamma distributions to represent the data and the statistical tests are based on these distributions. The second algorithm, called PolSeg, adopts the Wishart distribution as the model for the polanmetnc data. For the case of segmenting a single channel data this algorithm uses the test on equality of two Gamma parameters.