基于Penrose铺砌的数字图像重要性采样算法  

A Digital Image Importance Sampling Algorithm Based on Penrose Tiling

作　　者：姜玲燕 祝敏华 杨欢[3] JIANG Ling-yan;ZHU Min-hua;YANG Huan(Department of Physical Education,Suzhou Early Childhood Education College,Suzhou 215000,China;Research 52 of China Electronics Technology Group Corporation,Hangzhou 310000,China;School of Teacher Education,Nanjing Xiaozhuang University,Nanjing 211171,China)

机构地区：[1]苏州幼儿师范高等专科学校基础部,江苏苏州215000 [2]中国电子科技集团公司五十二研究所,浙江杭州310000 [3]南京晓庄学院教师教育学院,江苏南京211171

出　　处：《南京晓庄学院学报》2022年第6期113-118,共6页Journal of Nanjing Xiaozhuang University

基　　金：国家自然科学基金(61502137);江苏省高等教育教改研究(2021JSJG570).

摘　　要：重要性采样是数字图像处理领域的研究热点.在改进Penrose铺砌基本多边形的基础上,提出一种数字图像重要性采样算法.算法改进了Penrose铺砌的胖瘦菱形并利用其变换得到的6种多边形对图像进行非周期性铺砌,依据图像自身特征对非周期铺砌进行自适应细分,根据细分之后的图形产生相应的采样点,并通过控制细分程度来控制采样点的数量;为滤除采样产生的噪声,利用“斐波那契码”设定密度阈值,根据每个初始采样点的密度阈值判断该采样点的去留,从而降低采样产生的噪声;最后根据铺砌结构和原始图像的纹理重要性,创建一个8×21的二维数组(矫正向量表)来优化采样点,优化后的采样点集摆脱了Penrose铺砌的相对固定性,分布更加均匀.实验表明,该采样模型符合数字图像采样的光谱特性、精度高,符合原始图像特征.Importance sampling is an important and essential aspect of digital image processing.In this paper,we present a digital image importance sampling algorithm based on the improvement of Penrose tiling.In our algorithm,the image is aperiodic tiled by the six polygons which are transformed from the refined fat and thin diamond of Penrose tiling,adaptive subdivision of the aperiodic tiling is carried out according to the image features.The number of sampling points is controlled by controlling the degree of subdivision.In order to filter out the noise generated by sampling,Fibonacci code was used to set a density threshold for each sampling point,and judge whether to keep the sampling point or removed it as a noise point.Finally,we create an 8×21 two-dimensional array(correction scale)to optimize the sampling points which is according to theimportance of the pavement structure and the texture of the original image.The optimized sampling points get rid of the fixity of Penrose paving,and the distribution is more uniform.Experiments show that the sampling models have high precision andconform the spectral characteristic of digital image.

关 键 词：Penrose铺砌 自适应细分 矫正向量表 非周期铺砌 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]