Color Image Segmentation using Median Cut and Contourlet Transform : A Parallel Segmentation Approach

This paper presents a parallel implementation of color image segmentation algorithm using multiresolution technique. The idea is to achieve the complete and significant objects in the image using contourlet transform based image segmentation and to explore current multi-core architectures present in commercial processors in order to speed up the segmentation process for large size images. The algorithm attempts to use contourlet transform for texture feature extraction which can represent boundaries more accurately than that of discrete wavelet transform, so the contourlet transform is applied in proposed algorithm. This segmentation algorithm contains three stages. Initially the color quantization step is used to obtain a coarse image representation followed by parallel feature extraction using contourlet transform algorithm to exploit the multicore architecture. This task is executed sequentially by each assigned thread. Finally results returned by all threads are merged to achieve final segmentation output. Thus, color image segmentation can be obtained with reduced computational cost for large size images by proposed method. The algorithm is able to work with finite number of threads, so as to take full advantage of the upcoming processors having any number of cores. Keywords— Image segmentation, Color quantization, Contourlet transform, Region merging, Multicore processor