基于人眼视觉感知特性的自适应双边滤波算法  

An Adaptive Bilateral Filter Algorithm Based on Human Visual Perceptual Characteristics

作　　者：张有航 ZHANG You-Hang(BeiHang University Beijing100191,China)

机构地区：[1]北京航空航天大学机械工程及其自动化学院,北京100191

出　　处：《机电产品开发与创新》2022年第2期10-12,32,共4页Development & Innovation of Machinery & Electrical Products

摘　　要：针对图像去噪处理,双边滤波是一种比较典型的方法。它既考虑了参与滤波像素的距离因素,也考虑到了像素值差异,是一种边缘保护性滤波方法。但是,与传统图像去噪方法类似,双边滤波没考虑到图像不同区域内容特性,也没有考虑到人眼的视觉感知特性。因此,本文从人眼视觉角度出发,同时考虑图像内容的不同区域特性,采用了JND (Just Noticeable Difference最小可觉差)的视觉模型,提出一种符合人眼视觉感知特性的自适应双边滤波算法。实验结果表明,相比于传统双边滤波算法,我们的方法能够取得更好的去噪效果。Bilateral filtering is a typical method for image denoising. It takes into account both the distance factor of participating pixels and the difference of pixel values, and is an edge protective filtering method. However, similar to traditional image denoising methods, bilateral filtering does not consider the content characteristics of different regions of the image and the visual perception characteristics of human eyes.Therefore, in this paper, an adaptive bilateral filtering algorithm is proposed based on the visual model JND(Just Noticeable Difference),which is suitable for the visual perception characteristics of the human eye. Experimental results show that compared with traditional bilateral filtering algorithm, our method can achieve better denoising results.

关 键 词：双边滤波 块级滤波 人眼感知系统特性 JND 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]