Abstract: Aiming at the problem of over-segmentation in watershed algorithm, a new image segmentation algorithm based on bitmap cut and region merging is proposed. Firstly, the spatial gradient operator is used to obtain the initial gradient image, bitmap cut is designed to reconstruct the gradient image during the subsequent applications. Then, the watershed segmentation is performed on the new gradient image. Finally, the regions of the pre-segmentation result are merged that based on the principle of minimum het-erogeneity, and the final segmentation result is obtained. The proposed algorithm has been tested on different images and compared with other existing algorithms, the experimental results show the proposed algorithm is effective and efficient. Particularly, for the more challenging fuzzy edge, the accuracy of the proposed algorithm outperforms other algorithms.

Key words: image segmentation, bitmap cut, watershed, region merging

摘要： 就经典分水岭图像分割算法中存在的过分割问题，提出一种结合位图切割和区域合并的彩色图像分割算法。对原始彩色图像通过空域梯度算子求其梯度图像，并利用位图切割重建梯度图像；对新梯度图像进行分水岭预分割；对预分割图像基于异质性最小原则进行区域合并，并获得最终分割结果。相比于现有的同类方法，该算法引入位图切割，抑制噪声对分割结果的影响，在边缘模糊处分割准确，得到符合人类视觉的较小分割区域数目，同时在运行效率上提高。

关键词: 图像分割, 位图切割, 分水岭, 区域合并