基于聚类技术的多阈值图像分割技术

Multi-threshold image segmentation based on clustering method

多阈值图像分割中经常出现所选阈值数量不能事先确定的情况。为解决这一问题,提出一种基于聚类技术的多阈值图像分割方法,通过Mean Shift技术找出模式中心,通过迭代选择确定相邻模式中心的不同阈值,最后用多阈值完成对图像的分割。仿真实验结果表明,这一方法可有效应用于图像分割,并取得较好的效果。

The selected threshold quantity often cannot be determined in advance in image segmentation based on multi-threshold. A multi-threshold segmentation method is proposed based on clustering to solve the problem. The pattern centers are found through Mean Shift technology,various thresholds of adjacent pattern centers are determined by iterative selection method, and image segmentation is completed by the method of multi-threshold. The experimental results show that the proposed method is effective and can obtain good effect.