Automatic unsupervised texture segmentation using hidden Markov model

In this scheme, each texture is modeled as one HMM. Thus, if there are M different textures present in an image, there are M distinct HMMs to be found and trained. Consequently, the unsupervised texture segmentation problem becomes an HMM-based problem, where the appropriate number of HMMs, the associated model parameters, and the discrimination among the HMMs are the foci of the scheme. The scheme can be implemented by pipelined stages with no feedback from one stage to another, and each stage is highly suitable for parallel implementations. The scheme is evaluated using three textured images with different combinations of textures and is shown to perform with less than 3% error.<>