【摘要】 基于图像盲迭代复原,提出基于梯度的波纹评价方式、边缘轮廓变化评价波纹、平均边缘宽度评价波纹3种评价波纹的方法,得到一种新的基于梯度的波纹评价方式.此方法通过一个中心对称高斯加权窗在图像上逐点计算以该点为中心的邻域图像块的梯度相关信息;计算得到各点对应图像块的梯度信息并映射得到一幅图像来描述降质图像的降质信息,采用这幅降质信息图像的平均值作为整体图像质量的评价测度.对这3种图像质量评价方式进行实验发现,基于梯度评价方法可以很好地确定盲迭代的次数,从而得到相对质量较高的复原图像.更多还原

【Abstract】 Three methods were proposed based on blind iterative image restoration to evaluate the ripple:the gradient-based ripple assessment,ripple assessment with detecting the edge contour and with calculating an average width of the edge. A novel gradient-based ripple assessment approach was presented. The approach computed the gradient information within a centrosymmetry Gaussian weighted window,which moved pixel-by-pixel over the entire image. Then the information of corresponding image lumpy at all pixels was got,and an image was mapped to describe the degraded image. The average of the degraded image was adopted as the metric of image quality. The experiment of the three assessment methods shows that the gradient method can better determine the number of iteration and get restored images with better quality.更多还原