Automatic object extraction in nature scene based on visual saliency and super pixels
AICI'12 Proceedings of the 4th international conference on Artificial Intelligence and Computational Intelligence
| Hi-index | 0.00 |
To improve the performance of image segmentation, the paper suggests a new image segmentation method based on grey graph cut, which integrates grey theory and graph cut theory. In the method, the image is taken as a weighted undirected graph first. And then, after the relationships of grey-levels and positions in local regions are discussed via grey relational analysis, a grey weight matrix is established, based on which a grey partition function is constructed. Next, the image is binarized with the gray-level that corresponds to the minimum value of the grey partition function. Experimental results on visible light image and SAR image indicate that the proposed method, being superior to some existing methods like Otsu and Normalized Cut etc., not only can segment the images with obvious difference between targets and backgrounds, but also suppress image noise effectively.