Benchmarking of Objective Quality Metrics for Colorless Point Clouds

Abstract

Abstract-Recent advances in depth sensing and display technologies, along with the significant growth of interest for augmented and virtual reality applications, lay the foundation for the rapid evolution of applications that provide immersive experiences. In such applications, advanced content representations are required in order to increase the engagement of the user with the displayed imageries. Point clouds have emerged as a promising solution to this aim, due to their efficiency in capturing, storing, delivering and rendering of 3D immersive contents. As in any type of imaging, the evaluation of point clouds in terms of visual quality is essential. In this paper, benchmarking results of the state-of-the-art objective metrics in geometry-only point clouds are reported and analyzed under two different types of geometry degradations, namely Gaussian noise and octreebased compression. Human ratings obtained from two subjective experiments are used as the ground truth. Our results show that most objective quality metrics perform well in the presence of noise, whereas one particular method has high predictive power and outperforms the others after octree-based encoding.

Layered Projection-Based Quality Assessment of 3D Point Clouds

“…Considering that the objective scores heavily depend on the selected surface reconstruction algorithm. Thus, the po2mesh metric is considered as a sub-optimal solution for the quality assessment of PCs [4]. The po2point metric is based on geometric distances of associated points between the PC ref and PC dis , but it does not consider the fact that points in a PC usually represent surfaces on the object.…”

“…Depending on the capturing device, those formats can correspond to holograms [1], light fields [2], or point cloud (PC) [3], etc. Among these, PCs denote a practical content representation that allows users to visualize static or dynamic scenes in a more immersive way and can be directly exploited in systems aiming to provide immersive experiences with higher degrees of freedom [4], so PC recently becomes one of the first choices to represent 3D visual contents and has a wider variety of applications with the development of VR/AR [5], [6].…”

Be your own Benchmark: No-Reference Trajectory Metric on Registered Point Clouds

“…The main purpose of such metrics is to evaluate algorithm quality in the tasks of dense point cloud compression (down-sampling) and denoising. To evaluate metric quality, authors use collected benchmarks [19] from the subjective estimation of different point clouds' quality and estimate the correlation of proposed metrics using the Pearson correlation coefficient [20]. Among these approaches, the next metrics could be highlighted: p2point metric, p2plane metric [21], angular similarity [22], projection-based methods [23], [24] or 3D-SSIM variants [25], [26].…”

A comprehensive study of the rate-distortion performance in MPEG point cloud compression

“…Moreover, for the point-to-point and point-to-plane metrics, the geometric Peak-Signal-to-Noise-Ratio (PSNR) [26] is defined as the ratio of the maximum squared distance of nearest neighbors of the original content, potentially multiplied by a scalar, divided by the total squared error value, in order to account for differently scaled contents. The reader may refer to [23] for a benchmarking study of the aforementioned approaches. In the same category of geometry-only metrics falls a recent extension of the Mesh Structural Distortion Measure (MSDM), a well-known metric introduced for mesh models [34,41], namely PC-MSDM [27].…”