Fast Markovian Images Segmentation

This paper presents a solution to speed up the execution time of Hidden Markov Chain with Independent Noise model used in images segmentation. This solution is a traditional solution that divides the input image into a number of data blocks. Each block is treated independently to another following the same steps of estimation than the standard approach before dividing follows. The treatement of these blocks is sequentially. After the treatement of all blocks, we combine the result blocks to build the image result of segmentation. We compare our approach with the initial approach in term of complexity, segmentation quality and execution time. From the values of evaluated measures, we can confirm that our proposition provides the same results of segmentation than the initial approach but, our solution reduces the real execution time before dividing approximately by 40%.