CPC‐GSCT: Visual quality assessment for coloured point cloud based on geometric segmentation and colour transformation

Abstract

Coloured point cloud (CPC) is one of the important representations of three-dimensional objects, which has been used in many fields. CPC may encounter geometric and colour distortion during its compression, simplification or other processing. Thus, the objective visual quality assessment of CPC is one of the urgent issues to be resolved in the CPC's applications. Aiming at this problem, this paper proposes a new full-reference visual quality assessment metric for CPC based on geometric segmentation and colour transformation (CPC-GSCT), which analyzes geometric distortion and colour distortion of CPC. First, considering the visual masking effect of CPC's geometric information, CPC is segmented into different regions and distributed with different weights to describe the influence of visual masking effect in CPC quality assessment. At the same time, a geometric combination feature vector is defined and extracted for measuring the CPC's geometric distortion. Then, considering the colour perception of human eyes, a colour combination feature vector is extracted to measure the CPC's colour distortion in HSV colour space. Finally, all the extracted geometric and colour features are constituted as a feature vector to predict the quality of CPC. Experimental results on three databases (IRPC, SJTU-PCQA and CPCD2.0) show that the proposed CPC-GSCT metric can achieve better performance in predicting the visual quality of CPC than relevant existing methods.This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

Perceptual Quality Assessment of 3d Point Clouds

“…Later, MPEG point cloud database [64] and JPEG Pleno database [65] collect more point clouds from multiple resources, whose contents include cultural heritages, computer-generated objects, human figures, etc. The three point cloud databases provide a good number of visual stimuli for a series of subjective studies on point cloud quality [9], [13]- [15], [23], [24], [34], [36], [45], [48], [49], [52], [55], [68]. In fact, most point clouds used in existing publicly-available PCQA databases [9], [13]- [15], [23], [24], [34], [36], [45], [48], [49], [52], [55], [68] are selected from the three point cloud databases.…”

“…The three point cloud databases provide a good number of visual stimuli for a series of subjective studies on point cloud quality [9], [13]- [15], [23], [24], [34], [36], [45], [48], [49], [52], [55], [68]. In fact, most point clouds used in existing publicly-available PCQA databases [9], [13]- [15], [23], [24], [34], [36], [45], [48], [49], [52], [55], [68] are selected from the three point cloud databases. This makes existing PCQA databases suffer from similar shortcomings.…”

“…We believe that a large-scale subject-rated PCQA database of diverse, high quality source point clouds will benefit the PCQA research community. In this work, we first introduce our procedure of constructing a high quality Subject-rated point clouds IRPC [24] None, Color 6 PCL, G-PCC, V-PCC 54 vsenseVVDB [13] Color 2 V-PCC 32 vsenseVVDB2 [34] Color 8 Draco+JPEG, G-PCC, V-PCC 164 G-PCD [14], [15] None 5 Octree-puring, Gassian noise 40 RG-PCD [9] None 6 Octree-puring 24 M-PCCD [23] Color 8 G-PCC, V-PCC 244 PointXR [36] Color 5 G-PCC 100 NBU-PCD 1.0 [55] Color 10 Octree 160 CPCD 2.0 [49] Color 10 G-PCC, V-PCC, Gassian noise 360 SJTU-PCQA [45] Color 10 Octree, downsampling, color and geometry noise 420 ICIP2020 [48] Color 6 G-PCC, V-PCC 90 3DMDC [68] Color 5 QGeo, QCol, SGeo, SCol color point cloud, which comprises dense points, presents flawless object surfaces, and can be viewed from any direction. Following the same procedure, we build so far the largest source point cloud repository, consisting of 20 high quality point clouds with diverse geometric characteristics and textural patterns.…”

“…On the other hand, objective PCQA models are highly desired for practical use, as a subjective test is often expensive, labor-intensive and timeconsuming, restricting its usage in real-world applications. Recently, substantial effort has been made to develop objective PCQA models [12], [16], [21], [23], [25], [28]- [30], [32], [33], [35], [37]- [40], [44]- [47], [49]- [55], [55]- [60]. Existing models either calculate distortions by directly comparing the 3D structures between the reference and distorted point clouds [16], [25], [28]- [30], [33], [35], [37]- [40], [44], [46], [47], [49]- [51], [54], [55], [55]- [57], [59], [60], or leverage welldeveloped FR image quality assessment (IQA) models after projecting point clouds to multiple 2D images [12], [21], [23], [32], [45], [52], [53], [58].…”

“…Recently, substantial effort has been made to develop objective PCQA models [12], [16], [21], [23], [25], [28]- [30], [32], [33], [35], [37]- [40], [44]- [47], [49]- [55], [55]- [60]. Existing models either calculate distortions by directly comparing the 3D structures between the reference and distorted point clouds [16], [25], [28]- [30], [33], [35], [37]- [40], [44], [46], [47], [49]- [51], [54], [55], [55]- [57], [59], [60], or leverage welldeveloped FR image quality assessment (IQA) models after projecting point clouds to multiple 2D images [12], [21], [23], [32], [45], [52], [53], [58]. However, the performances of these models are often degraded by the absence of proper modelings of the HVS [69],…”