一种图像降噪的自适应迭代终止策略  

AN ADAPTIVE ITERATION TERMINATION STRATEGY FOR IMAGE DENOISING

作　　者：戴蓉[1] 黄成[2] 

机构地区：[1]中国民用航空飞行学院计算机学院,四川广汉618307 [2]四川工程职业技术学院,四川德阳618000

出　　处：《计算机应用与软件》2016年第5期204-206,238,共4页Computer Applications and Software

基　　金：四川省教育厅科研项目(13ZA0135)

摘　　要：降噪是图像处理领域的重要问题。针对现有大多数迭代类图像降噪算法终止条件判断复杂或不恰当终止的现象,提出一种与图像相关的迭代终止判据,不需要预设经验参数,而通过对图像数据的学习实现自适应的方法。该方法通过判断每一步被降噪的像素点的数量是否最小化来决定是否终止迭代,从而使处理后的图像具有最大峰值信噪比。实验结果表明,该方法在各种噪声条件图片的降噪中都实现了峰值信噪比最大化,是一种高效的、与图像相关的方法。Noise reduction is an important issue in the field of image processing. For the phenomena that most of current image noise reduction algorithms of iteration class are complex in judging the termination condition or are inappropriately terminated,we proposed such a method,it has the iteration termination criterion correlated to images and needs no preset empirical parameters but achieves the adaption through learning image data. The method decides whether to terminate the iteration or not according to the judgement of whether the number of the denoised pixels in each step be minimised,so that the processed image will have a maximum peak signal-to-noise ratio（ PSNR）.Experimental result revealed that the method achieved the maximisation of PSNR in image noise reductions with various noisy environments,and was a highly efficient and image-correlated method.

关 键 词：降噪 迭代终止条件 自适应 峰值信噪比 选择性中值滤波器 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]