This study presents a novel approach to reduce over-segmentation using both pre- and post-processing for watershed segmentation. We make use of more prior knowledge in pre-processing and merge the redundant minimal regions in post-processing. In the initial stage of the watershed transform, this not only produces a gradient image from the original image, but also introduces the texture gradient. The texture gradient can be extracted using a gray-level co-occurrence matrix. Then, both gradient images are fused to give the final gradient image. After the initial results of segmentation, we use the merging region technique to remove small regions. Experiments show the effectiveness of segmentation.
展开▼
