摘要
将前向后向扩散系数引入到You和Kaveh提出的四阶偏微分方程去噪模型中,前向扩散用于对噪声进行平滑,后向扩散则对图像特征进行强化。同时,改进了模型中拉普拉斯算子的离散形式,使其包含更多的图像信息,能够更准确的判断图像的特征。新方法处理后的图像,避免了二阶偏微分方程处理图像常出现的"阶梯"效应,同时,和同类的四阶偏微分方程去噪模型相比,该方法的处理结果不会出现"斑"点,因此视觉效果更加理想。最后,通过实验证明了该方法的有效性。
In this paper,the forward and backward diffusion coefficients are introduced to the fourth-order partial differential equation which was proposed by You and Kaveh for removing the noise.The forward diffusion is used to smooth the noise,and the backward diffusion is used to enhance the image characteristics.Meanwhile,the discretization scheme of Laplace operator is improved,and more image information can be contained.Thus,the image features can be judged more accurately.The image processed by the proposed model do not occur the blocky effects which are widely seen in images processed by second-order nonlinear diffusion.And compared to other fourth-order partial differential equations,the image processed by the proposed model do not generate the speckles,so the visual effects are superior to the image processed by You-Kaveh model.Finally,the validity of the proposed model is proved by experiments.
出处
《计算机与数字工程》
2010年第12期108-111,共4页
Computer & Digital Engineering
