Modeling and Segmentation of Noisy and Textured Images Using Gibbs Random Fields
IEEE Transactions on Pattern Analysis and Machine Intelligence
| Hi-index | 0.00 |
This paper proposes the use of the inherent characteristics of SAR images to improve Gibbs-MRF model for recovering SAR image. Further, it puts forward to segment SAR image into target and shadow with the theory of connectivity in digital morphology. The new method is not only using GAMMA distribution to replace the traditional Rayleigh distribution in the estimate of MAP (Maximum A Posteriori Probability, MAP), but also using the connectivity model of pixels intensity value relevance to extract goal better in the neighborhood of SAR image pixel space. This method takes full advantage of the relevance between the information of digital morphology of the SAR image and the pixel intense, and eliminates isolated points and obtains good segment results.