The researches of the image segmentation algorithm based on Contourlet transformation

The image Segmentation is resulted, by Wavelet Domain Hidden Markov Tree model, dim edge components prone and diffuse singularity. The Contourlet transformation can capture the singularity in the image of high dimension, therefore, the paper presents a new algorithm of multiscale image segmentation based on the Contourlet transform. This algorithm was initially segmented in scales by Contourlet domain Hidden Markov Tree Model, and integrated among multiscales after segmentting images through the adaptive context structures. Compared simulating segmentation experiment of synthetic texture image and aerial image to Wavelet Domain Hidden Markov Tree Model image segmentation method, we obtain more desirable segmentation effect with the regional consistency and edge accuracy, and the composed images with lower misclassification rate.