利用非线性扩散的半自动纹理图像分割

基金项目: 国家科技部科研院所社会公益研究分项资助项目(2004DIB2J067)

详细信息

Semi-automatic Segmentation of Texture Image Based on Nonlinear Diffusion

¹ Spatial Information Technology Application Research Institute,Yangtze River Scientific Research Institute,23 Huangpu Street,Wuhan 430010,China

Funds: 国家科技部科研院所社会公益研究分项资助项目(2004DIB2J067)

摘要: 介绍了一种非线性扩散过滤器法从包含纹理的图像中获取灰度、尺度和方向信息,并形成经过耦合保边平滑的5通道向量值图像,对纹理图像的分割变为对此向量值图像的分割。针对半自动操作的特点,在合理的假设前提下,采用了多通道统计区域分割法。试验结果表明,本文的纹理分割方法能有效地利用重要纹理特征与灰度的混合信息,是一种半自动的无监督纹理图像分割方法。
- 纹理分割 /
- 非线性扩散 /
- 区域统计动态轮廓
Abstract: A Texture image segmentation method based on nonlinear diffusion is presented.Multi-band statistical region active contour is used to segment the vector image.The method can be seen as a unsupervised one because parameters are not sensitive to the different texture images.Experimental results show that the method is effective.
- texture image segmentation /
- nonlinear diffusion /
- statistical region active contour