In this work, a new approach to color image segmentation is presented. First, three segmentations on 2D color space RG, RB and GB are obtained respectively. Then, these 2D segmentations are combined to obtain a final segmentation using fuzzy sets and image fusion, which is implemented by a region splitting-Merging process. Experiments show that this approach provides improved performance, especially in running time and memory consumption.
展开▼