基于OLS算子和小波分解的分水岭图像分割算法

An improved watershed segmentation algorithm based on OLS operator and wavelet decomposition

针对分水岭图像分割算法导致目标图像被过分割或者欠分割的不足,提出了一种基于OLS算子和小波分解的改进分水岭算法。首先,利用OLS算子和小波分解构造多尺度的边缘保持方法。然后,通过增加或减少层次,在保证平滑性和细节性基础上选择最佳新图像。最后,基于分水岭分割算法对选择的图像进行分割。试验结果表明,该方法能够更准确地分割图像,使用的技术比其他算法分割更加精确,对于复杂图像,其分割精度维持在0.80以上。

In order to solve defects as over segmentation or under segmentation of target image in current watershed image segmentation algorithm, an improved watershed algorithm based on OLS operator and wavelet decomposition is proposed. First, multiscale edge preserving methods is constructed by using OLS operator and wavelet decomposition. Second, the best new image is selected by increasing or decreasing some levels on the basic of ensuring smoothness and detail. Third, segmentation of selected images is based on watershed segmentation algorithm. The experimental results show that the proposed method can segment the image more accurately, its segmentation accuracy in complex scenes can reach 0.8, which has a good reference value in pattern recognition and image processing.