Application of Gaussian Super-Pixels Based Quick Graph Cuts in Image Segmentation

Abstract

The fusion of the edge confidence and Mean Shift algorithm is efficiently segmented of the original image for more accurate boundary homogeneous regions, and these regions are described as super-pixel. It is been used to build streamlined weighted graph. The paper has presented an interactive image segmentation algorithm. It is constructed by using Gaussian super-pixels and quick graph cuts model to achieve algorithm’s acceleration. Then, the uses of regional color statistic features are described in the super-pixel and information theory space Gaussian maximize the distance metric design. The expectation of the order parameter is learning on a priori knowledge. Accurate and concise data is also used in component form mixed Gaussian algorithm for clustering interaction. Finally, the paper has proposed the improved weighted figure model with quick graph cuts algorithm to obtain the better segmentation result than others. The image segmentation experiments by using a different color image comparison. The simulation results are shown that the proposed method has good performance in the accuracy and efficiency.