【摘要】 为了增强图像边缘和纹理以及去除图像噪声,讨论分数阶微积分算子频域特性,提出新型的分数阶微积分算子模板和算法.考虑边缘和纹理具有方向性、空间邻域内像素的关联性以及相邻像素灰度值的相近性,利用3×3模板,在分数阶微分增强时,取中心点4个方向最大值为增强值;在分数阶积分去噪时,将图像分别在8个方向卷积,按照在8个方向卷积结果总和中所占的比例大小进行线性加权,从而得到去噪后的图像.实验结果表明,提出的分数阶微积分算子能明显地增强图像的边缘和纹理信息以及去除图像噪声,并且视觉效果明显.更多还原

【Abstract】 In order to improve the ability of image enhancement and denoising,the frequent domain characteristics of fractional calculus are discussed,and some new improved fractional calculus masks and methods are proposed. The edge and texture of images have direction,and there is spatial correlation between the neighbored pixels,so there is the adjacent gray value. When the image is enhanced using 3×3 template of fractional differential operator,the enhancing value,which is maximum value of the four direction values,can be gained. Taking the arithmetic mean of the weighting sum value in 3×3 template as approximate value of denoised image,a novel image denoising approach based on two improved fractional integral masks is also proposed. The experimental results of Lena and Camera show that the improved method can effectively enhance image edge and texture information and remove noise respectively. Furthermore,its advantages include good visual effects.更多还原