基于全局与局部感知特征的三维彩色点云质量评价  

Exploiting global and local perceptual features for quality assessment of 3D colored point cloud

作　　者：何周燕 蒋志迪[1] 郁梅[1] HE Zhouyan;JIANG Zhidi;YU Mei(Faculty of Information Science and Engineering,Ningbo University,Ningbo,Zhejiang 315211,China)

机构地区：[1]宁波大学信息科学与工程学院,浙江宁波315211

出　　处：《光电子．激光》2021年第10期1046-1054,共9页Journal of Optoelectronics·Laser

基　　金：国家自然科学基金项目(62071266)资助项目。

摘　　要：作为物理对象在三维空间的有效表示方法,三维彩色点云可以提供丰富的沉浸式视觉体验,但在其获取、处理、编码传输等各环节会引入失真,从而导致其视觉质量下降。因此,如何监测彩色点云的视觉质量是一个亟待解决的重要问题。本文将三维彩色点云投影到二维平面,提出了一种基于全局与局部感知特征的彩色点云视觉质量评价方法。首先,将三维彩色点云转化为彩色纹理投影图与几何投影图。然后,根据三维彩色点云的纹理与几何失真在其投影图中的不同表象,分别描述并提取其失真特征;其中,在彩色纹理投影图中提取全局颜色与局部纹理特征,在几何投影图中提取全局与局部几何特征。最后,将所有全局和局部感知特征构成最终的特征向量预测彩色点云的视觉质量。在两个主观评价数据库(SJTU-PCQA、CPCD2.0)进行测试的实验结果表明,所提出方法在性能上优于13个现有代表性视觉质量评价方法,与主观感知质量有更好的一致性。As an effective representation technology for physical objects in three-dimensional(3 D)space,3 D colored point cloud(PC)can provide users with rich immersive visual experience.However,distortions will be introduced in acquisition,processing,coding and transmission of 3 D colored PC,resulting in the decline of visual quality.Therefore,how to monitor the visual quality of 3 D colored PC is an important problem to be solved.By projecting 3 D colored PC onto two-dimensional planes,this paper proposes a novel visual quality assessment method for colored PC based on global and local perception features.Firstly,3 D colored PC is converted into color texture projection map(CTPM)and geometric projection map(GPM).Then,feature extraction is performed according to different representations of texture and geometric distortions,including global color and local texture features based on CTPM,and global and local geometry features based on GPM.Finally,all the global and local quality perception features constitute the final feature vector to predict the visual quality of colored PC.Experimental results on two subjective quality assessment databases(SJTU-PCQA,CPCD2.0)show that the performance of the proposed method is superior to that of thirteen existing representative visual quality assessment methods,and it is more consistent with the subjective quality.

关 键 词：彩色点云 质量评价 全局与局部 感知特征 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]