改进的Mean Shift滤波在砂砾岩图像分割中的应用

砂砾岩图像成分颜色多样、色彩不均,要有效地进行分割,提取砾岩颗粒,图像预处理就显得尤为重要。MeanShift滤波相比传统的滤波算法具有更好的平滑特性,能很好地消除噪声和色彩不均对后续分割造成的影响,但运算速度较慢。针对该问题,提出利用高斯金字塔和像素点赋值方法改进Mean Shift滤波算法速度。采用改进后的Mean Shift滤波算法对砂砾岩图像进行预处理,然后进行分割,提取粒岩,取得了较好的效果。

Application of Improved Mean Shift Wave Filtering in Segmentation of Glutenite Image

The ingredients of glutenite images have diversity and uneven colors.In order to segment effectively,it is very important to preprocess the images before refining the glutenite kernel.Compared with the traditional wave filtering arithmetic,Mean Shift wave filtering have a better smoothness characteristic.It is good for eliminating the noises and uneven colors which would have bad effect on the subsequent segmentation.However,the speed of arithmetic is too slow.To the question,Gauss Pyramid and Pixel Evaluate was applied to improve the arithmetic of Mean Shift wave filtering.Before the step of segmentation and extraction of granulite,we could adopt the improved Mean Shift wave filtering arithmetic to preprocess the glutenite images,which can grantee the speed of process and have a better effect.