Towards a Point Cloud Structural Similarity Metric

Abstract

Point cloud is a 3D image representation that has recently emerged as a viable approach for advanced content modality in modern communication systems. In view of its wide adoption, quality evaluation metrics are essential. In this paper, we propose and assess a family of statistical dispersion measurements for the prediction of perceptual degradations. The employed features characterize local distributions of point cloud attributes reflecting topology and color. After associating local regions between a reference and a distorted model, the corresponding feature values are compared. The visual quality of a distorted model is then predicted by error pooling across individual quality scores obtained per region. The extracted features aim at capturing local changes, similarly to the wellknown Structural Similarity Index. Benchmarking results using available datasets reveal best-performing attributes and features, under different neighborhood sizes. Finally, point cloud voxelization is examined as part of the process, improving the prediction accuracy under certain conditions.

A Reduced Reference Metric for Visual Quality Evaluation of Point Cloud Contents

“…Viola et al incorporate color distortion in geometrybased metrics, using luminance histogram information [10], whereas Diniz et al use local binary pattern descriptors to estimate texture distortion [11]. In [12], Alexiou et al propose the usage of local statistical features in order to obtain a global measure of degradation, similarly to the Structural Similarity Index (SSIM) in the image domain. Projection-based metrics rely on mapping the original and distorted point clouds on planar surfaces, and then using popular image quality assessment metrics on the resulting projected images.…”

“…The dataset was created by applying distortions uniquely on the geometry domain, while the color information was uncompressed, and obtained through recoloring. Thus, the color information may act as a distractor [12], hiding impairments in the geometry domain. As our proposed weights heavily include a measure of color distortion, a less than optimal performance in this dataset is to be expected.…”

Visual Quality of Compressed Mesh and Point Cloud Sequences

“…In [52], a plane-to-plane FR metric measured angular similarity through the intersection angle of normal vectors between two corresponding points. While the aforementioned work also focused only on geometric distortion, two studies considered color attributes as well [53], [54]. In [53], the authors rendered a 3D point cloud onto a 2D plane, then applied traditional image FR quality metrics to measure the 2D image quality as input to their prediction model for perceptual quality of the 3D PC.…”

“…In [53], the authors rendered a 3D point cloud onto a 2D plane, then applied traditional image FR quality metrics to measure the 2D image quality as input to their prediction model for perceptual quality of the 3D PC. Different objective metrics were proposed in [54] based on geometry, normal vectors, curvature and color separately, and the color-based metric was found to best match perceptual quality.…”

Layered Projection-Based Quality Assessment of 3D Point Clouds

“…Objective PCQA metrics are commonly classified as Full Reference (FR) [17]- [28], Reduced Reference (RR) [29], [30], and No Reference (NR), depending on the avail- ability of reference information. Compared with RR and NR metrics, people have more research on FR metrics which can be classified as (a) point-based [18]- [25] and (b) projection-based [17], [26]- [28]. Most studies in the past had emphasized on the geometric distortion measurement of PC object, such as the point-to-point (po2point), point-toplane (po2plane), point-to-mesh (po2mesh) [18] and planeto-plane (pl2plane) [19].…”