【摘要】 基于图像在切线方向和法线方向上的不同扩散要求,分别设定了不同的扩散系数.在切线方向上,图像中各个位置都要求大的扩散作用以有效去除噪声,故设定扩散系数恒为1.在法线方向上,内部和边缘具有不同的扩散要求,可利用数学形态学的方法估计每步迭代扩散后即时图像的梯度模阈值,用自适应阈值取代Perona-Malik扩散系数中的固定阈值,从而得到适合即时图像的法向扩散系数,目的是有效区分内部区域和边缘以确保尽可能多地保留边缘信息.实验表明,该方法在细小边缘信息的保留方面具有较明显的效果.更多还原

【Abstract】 According to the different diffusion demand of image in tangential and normal direction,the different diffusion parameter is set.In the tangential direction,the diffusion should be large enough to remove noise efficiently,so the diffusion parameter is 1.In the normal direction,the diffusion demand is different in the interior and edge of an image.We control the threshold of gradient modulus by the morphology method.After substituting the stable threshold by this adaptive threshold in Perona-Malik diffusion parameter,the suitable normal diffusion parameter for the instant image is obtained.The intention is that the interior and edge can be distinguished effectively and the edge information can be preserved as much as possible.Experimental results show that our method has an obvious effect in preserving small details.更多还原