基于局部特征核估计的图像去模糊算法

Algorithm of image deblurring based on local features kernel estimation

针对在图像去模糊迭代处理阶段的潜像估计和模糊核估计处理速度较低的问题,采用了一种从中等尺寸的单一运动模糊图像实现去模糊的方法,该方法利用简单的图像处理技术从一张预估计的潜像中预测出强边缘,然后得到的结果又独立地运用到模糊核的估计中,能很好地抑制在预测阶段反卷积形成的不良效应,同时构成的数字化系统展现出来的结果比使用像素值计算的需要更少的条件限制,结果收敛得更快。仿真实验结果表明:在去模糊质量相当的情况下,此方法比以往成果的处理速度快20%。

According to the fuzzy iterative processing stage in the image of the latent image estimation and fuzzy kernel estimation low processing speed,this paper presents a motion from a single medium size to achieve fuzzy image deblurring method. This method uses simple image processing techniques from a pre-estimation of the latent image in the prediction of strong edges, then the results obtained independently using to estimate the blur kernel,can well inhibit the formation of adverse effects in the prediction stage of disconsolation. At the same time,the results of the digital system show that the result of the digital system is much less than that of the pixel value,and the result of the digital system converges more quickly. The simulation results show that this method is faster 20% than that of the previous results in the case of fuzzy quality.