45N度角旋转梯度等模扩散控制图像滤波方法

Diffusion Control Image Filtering Method of 45N Degree Rotation Invariant Gradient Magnitude

在基于偏微分方程的图像去噪方法中,存在扩散控制力较弱的问题,导致图像边缘在扩散处理中形状保持能力下降,影响去噪效果,为了更好地解决这一问题,提出一种45N度角旋转梯度等模扩散控制图像滤波方法。首先,分析了现有扩散方程的不足,即梯度计算与扩散控制。其次,通过引入旋转角计算邻域四角梯度矢量在x、y轴方向的投影,并与水平及垂直矢量叠加,获得一种在45N度上具有梯度模值不变性的矢量计算模板。最后,采用该模板分别计算x与y方向的一阶偏导数,并以此统计梯度矢量方向一致性,改进边缘停止函数,获得更好地扩散控制。实验结果验证了本文算法的有效性。

Problem of weaker diffusion control leads to decline the edge shape of retention capacity based on PDE of im-age denoising,and the results should be poor.For this reason,the paper proposes a diffusion control image filtering method of 45 N degree rotation invariant gradient magnitude.Firstly,shortcomings of the diffusion equation are analyzed,they are gradient calculation and diffusion control.Secondly,the invariant vector gradient template is adopt through this method by introducing the rotation angle calculation of four neighborhood angle projection gradient vector in 45 N degree.x and y direc-tion of the first order partial derivative is calculated using this template,and the consistency of gradient vector direction is counted,so the edge stopping function is adopt to obtain better diffusion control.The results of experiment show that algo-rithm has validity.