一种基于WMF-ACA的深度图像修复算法  

A depth image restoration algorithm based onWMF-ACA

作　　者：沈慧中 张倩[1] 王斌[1] 孙艺晋 吴玉静[2] 王晓慧 SHEN Huizhong;ZHANG Qian;WANG Bin;SUN Yijin;WU Yujing;WANG Xiaohui(College of Information,Mechanical and Electrical Engineering,Shanghai Normal University,Shanghai 201418,China;High School Affiliated to Fudan University,Shanghai 200433,China;Shanghai Municipal Educational Examinations Authority,Shanghai 200433,China)

机构地区：[1]上海师范大学信息与机电工程学院,上海201418 [2]复旦大学附属中学,上海200433 [3]上海市教育考试院,上海200433

出　　处：《上海师范大学学报（自然科学版）》2021年第4期418-422,共5页Journal of Shanghai Normal University(Natural Sciences)

摘　　要：针对RGBD相机由于硬件条件的限制,所获取的深度图像往往存在空洞与噪声的问题,提出一种彩色图像引导的深度图像空洞填补方法.对深度图像边缘区域采用基于局部直方图的加权模式滤波器(WMF)进行处理,在有效保留深度图边缘与细节的前提下,消除图像噪声.对深度图像非边缘区域采用一种异步元胞自动机(ACA)模型算法,根据邻域和迭代规则对深度图空洞进行填补,快速、准确地完成任务.在立体匹配数据集Middlebury上测试所提算法实现性能,实验结果表明:该算法的结构相似性(SSIM)与峰值信噪比(PSNR)更高,能够准确地填补深度图空洞,同时保持深度图边缘清晰.To solve the problem of holes and noise in the depth image due to the limitation of RGBD camera hardware,a color imageguided method for filling holes in depth image was proposed.Firstly,for the boundary area of the depth image,a weighted mode filter(WMF)based on local histogram was used for processing which effectively preserved the boundary and details and eliminated noise of the depth image.Secondly,an asynchronous cellular automaton(ACA)model algorithm was adopted for the non-boundary area of the depth image.The depth map holes were filled by using the neighborhood and iteration rules and the mission was completed fleetly and accurately.Finally,the performance of the proposed algorithm was tested on Middlebury.Experimental results showed that the proposed algorithm in this paper could accurately fill the holes in the depth map with a clear boundary.

关 键 词：深度图 空洞填补 加权模式滤波器(WMF) 异步元胞自动机(ACA) 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]