基于改进的自适应分水岭图像分割方法研究

Based on the Improved Adaptive Watershed Image Segmentation Method Research

研究图像分割中由于对噪声的抑制能力弱以及对大多数图像易产生过分割现象,从而导致图像分割过程中局部分割线产生偏移现象,进而使得图像分割变得困难。为解决上述问题,提出了改进的自适应分水岭图像分割方法。首先,对输入的图像进行自适应降噪滤波,以减弱因噪声干扰导致的区域极小值;然后,运用形态梯度算子对滤波去噪后的图像进行平滑处理,以减弱噪声对分水岭分割的影响;最后,对图像进行目标标记,用来屏蔽消除其它无用的极小值,仅允许标记过的极小值生长为分割区域并得到最终的分割线,达到最终分割出感兴趣物体的目的,仿真结果表明,与传统分水岭分割方法比较,缓解了分水岭算法过分割问题,增强分割算法的鲁棒性,优于一般的分水岭算法。

Of image segmentation due to noise interference suppression capacity is weak and easy to produce the majority of image segmentation phenomenon, leading to the local split line in the image segmentation process offset phenomenon, thus making segmentation difficult. In view of this, this paper presents an improved adaptive watershed image segmentation method. First, the input image -adaptive noise reduction filter, in order to weaken due to noise caused by regional minimum; then use the morphological gradient operator smoothing filter denoising images, to weaken the noise on the watershed segmentation goal mark; Finally, the image used to shield eliminate unwanted minimum, allowing only the minimum growth mark objects of interest is the partition and a final split line, the final split simulation results show that, compared with the traditional watershed segmentation method to alleviate the over - segmentation of watershed algorithm, and enhance the robustness of the segmentation algorithm. Superior to the watershed algorithm.