基于融合自适应形态滤波的分水岭分割新算法*

针对分水岭分割易产生过分割问题，提出一种基于融合自适应形态滤波的分水岭分割算法。该算法对图像进行多结构多尺度自适应形态滤波处理，从而抑制图像的暗噪声和暗纹理细节。为了增强目标和背景的对比度，进行高低帽变换，再应用浸没模型的分水岭算法进行图像分割。实验结果表明，与一般的分割算法相比，该算法能有效地抑制各种噪声对分割的影响，具有较强的抗过分割性能，快速有效地实现图像的分割。

Improved algorithm of watershed segmentation based on fusion and adaptive morphological filtering

To overcome the over-segmentation for the watershed, this paper presented a novel algorithm of watershed segmentation based on fusion and adaptive morphological filtering. The algorithm first executed the adaptive morphological filtering of fusion to restrain dark noise and texture details of the images. Secondly, in order to enhance the contrast of the object and background, carried out the Tophat and Bothat transformations. Finally, achieved the image segmentation using the watershed algorithm based on immersion model. The experimental results show that the proposed algorithm, compared with general segmentation algorithms, can effectively overcome the segmentation impact on all kinds of noise, has a strong performance against over-segmentation, and achieves quickly and efficiently the image segmentation.