Adaptive Downsampling and Spatial Upconversion for Point Cloud Compression

The integration of three-dimensional (3D) data defined in different coordinate systems requires the use of well-known registration procedures, which aim to align multiple models relative to a common reference frame. Depending on the achieved accuracy of the estimated transformation parameters, the existing registration procedures are classified as either coarse or fine registration. Coarse registration is typically used to establish a rough alignment between the involved point clouds. Fine registration starts from coarsely aligned point clouds to achieve more precise alignment of the involved datasets. In practice, the acquired/derived point clouds from laser scanning and image-based dense matching techniques usually include an excessive number of points. Fine registration of huge datasets is time-consuming and sometimes difficult to accomplish in a reasonable timeframe. To address this challenge, this paper introduces two down-sampling approaches, which aim to improve the efficiency and accuracy of the iterative closest patch (ICPatch)-based fine registration. The first approach is based on a planar-based adaptive down-sampling strategy to remove redundant points in areas with high point density while keeping the points in lower density regions. The second approach starts with the derivation of the surface normals for the constituents of a given point cloud using their local neighborhoods, which are then represented on a Gaussian sphere. Down-sampling is ultimately achieved by removing the points from the detected peaks in the Gaussian sphere. Experiments were conducted using both simulated and real datasets to verify the feasibility of the proposed down-sampling approaches for providing reliable transformation parameters. Derived experimental results have demonstrated that for most of the registration cases, in which the points are obtained from various mapping platforms (e.g., mobile/static laser scanner or aerial photogrammetry), the first proposed down-sampling approach (i.e., adaptive down-sampling approach) was capable of exceeding the performance of the traditional approaches, which utilize either the original or randomly down-sampled points, in terms of providing smaller Root Mean Square Errors (RMSE) values and a faster convergence rate. However, for some challenging cases, in which the acquired point cloud only has limited geometric constraints, the Gaussian sphere-based approach was capable of providing superior performance as it preserves some critical points for the accurate estimation of the transformation parameters relating the involved point clouds.

Planar-Based Adaptive Down-Sampling of Point Clouds

The perturbation features limit the performance of point cloud classification models both for clean point cloud and point cloud obtained from real-world. In this paper, we propose two methods to enhance models by reducing the negative impact of the nuisance features from the perspective of interpretability, namely, dropping the nuisance points before inputting the point cloud into models and adaptively selecting the important features during the training process. The former is achieved by saliency analysis of models, the perturbation points are those with low contribution to models. For each sample, dropping the low contribution points based on the saliency scores is equivalent to filtering the perturbation features. We design a generic framework for generating saliency maps for various models and datasets, and obtain empirical values for the number of dropped points on each set of them. Then we apply the unsupervised dropping process to improve the robustness of models. The latter is achieved by adaptive downsampling, and we design a multi-stage learnable class-attention-based downsampling module to replace the commonly used Farthest Point Sampling (FPS). As the training progresses, the downsampling module tends to select the common features for each category, thus eliminating the nuisance features to improve the learning efficiency of the model. For dropping points (DP), we generate saliency maps for PointNet&++, DGCNN and PointMLP on ModelNet40 and ScanObjectNN, PointNet+DP reaches an overall accuracy (OA) of 92.5% and 72% on ModelNet40 and ScanObjectNN, surpassing the original model by 3.4% and 5.3%, the OA of PointNet++ with DP on ModelNet40 object classification can be raised from 91.8% to 93.7%. For adaptive feature selection (AFS), PointMLP-elite+AFS reaches an OA of 92.5% and a mean accuracy (mAcc) of 72% on ScanObjectNN, surpassing the original model by 0.8% and 1%. It reaches the level of PointMLP with 6.3% of the number of parameters. Considering the difficulty of the deployment of deep models, PointMLP-elite+AFS is the most cost-effective classification model known on ScanObjectNN.

Point cloud acquisition techniques are an essential tool for the digitization of industrial plants, yet the bulk of a designer's work remains manual. A first step to automatize drawing generation is to extract the semantics of the point cloud. Towards this goal, we investigate the use of deep learning to semantically segment oil and gas industrial scenes. We focus on domain characteristics such as high variation of object size, increased concavity and lack of annotated data, which hampers the use of conventional approaches. To address these issues, we advocate the use of synthetic data, adaptive downsampling and context sharing.

This meta-survey provides a comprehensive review of 3D point cloud (PC) applications in remote sensing (RS), essential datasets available for research and development purposes, and state-of-the-art point cloud compression methods. It offers a comprehensive exploration of the diverse applications of point clouds in remote sensing, including specialized tasks within the field, precision agriculture-focused applications, and broader general uses. Furthermore, datasets that are commonly used in remote-sensing-related research and development tasks are surveyed, including urban, outdoor, and indoor environment datasets; vehicle-related datasets; object datasets; agriculture-related datasets; and other more specialized datasets. Due to their importance in practical applications, this article also surveys point cloud compression technologies from widely used tree- and projection-based methods to more recent deep learning (DL)-based technologies. This study synthesizes insights from previous reviews and original research to identify emerging trends, challenges, and opportunities, serving as a valuable resource for advancing the use of point clouds in remote sensing.

A point cloud is a set of 3D points that can be used to represent a 3D surface. Each point has a spatial position (x, y, z) and a vector of attributes, such as colors, material reflection, or normal. As point clouds are capable of reconstructing 3D objects or scenes, they have the potential to be widely used in various applications such as auto-driving and 6-degree virtual reality. However, the following properties of point cloud make the point cloud compression and processing become rather challenging. 1) Unstructured. The point cloud is a series of non-uniform sampled points. On the one hand, it makes the correlations among various points difficult to be utilized for compression. On the other hand, the convolutional neural network that is widely used in image/video processing cannot be applied to the point cloud processing. 2) Unordered. Unlike images and videos, the point cloud is a set of points without a specific order. Therefore, both the point cloud processing and compression algorithms need to be invariant to any permutations of the input point clouds.

With the growth of Extended Reality (XR) and capturing devices, point cloud representation has become attractive to academics and industry. Point Cloud Compression (PCC) algorithms further promote numerous XR applications that may change our daily life. However, in the literature, PCC algorithms are often evaluated with heterogeneous datasets, metrics, and parameters, making the results hard to interpret. In this article, we propose an open-source benchmark platform called PCC Arena. Our platform is modularized in three aspects: PCC algorithms, point cloud datasets, and performance metrics. Users can easily extend PCC Arena in each aspect to fulfill the requirements of their experiments. To show the effectiveness of PCC Arena, we integrate seven PCC algorithms into PCC Arena along with six point cloud datasets. We then compare the algorithms on ten carefully selected metrics to evaluate the quality of the output point clouds. We further conduct a user study to quantify the user-perceived quality of rendered images that are produced by different PCC algorithms. Several novel insights are revealed in our comparison: (i) Signal Processing (SP)-based PCC algorithms are stable for different usage scenarios, but the trade-offs between coding efficiency and quality should be carefully addressed, (ii) Neural Network (NN)-based PCC algorithms have the potential to consume lower bitrates yet provide similar results to SP-based algorithms, (iii) NN-based PCC algorithms may generate artifacts and suffer from long running time, and (iv) NN-based PCC algorithms are worth more in-depth studies as the recently proposed NN-based PCC algorithms improve the quality and running time. We believe that PCC Arena can play an essential role in allowing engineers and researchers to better interpret and compare the performance of future PCC algorithms.

With the maturity of 3D capture technology, the explosive growth of point cloud data has burdened the storage and transmission process. Traditional hybrid point cloud compression (PCC) tools relying on handcrafted priors have limited compression performance and are increasingly weak in addressing the burden induced by data growth. Recently, deep learning-based PCC methods have been introduced to continue to push the PCC performance boundary. With the thriving of deep PCC, the community urgently demands a systematic overview to conclude the past progress and present future research directions. In this paper, we have a detailed review that covers popular point cloud datasets, algorithm evolution, benchmarking analysis, and future trends. Concretely, we first introduce several widely-used PCC datasets according to their major properties. Then the algorithm evolution of existing studies on deep PCC, including lossy ones and lossless ones proposed for various point cloud types, is reviewed. Apart from academic studies, we also investigate the development of relevant international standards (i.e., MPEG standards and JPEG standards). To help have an in-depth understanding of the advance of deep PCC, we select a representative set of methods and conduct extensive experiments on multiple datasets. Comprehensive benchmarking comparisons and analysis reveal the pros and cons of previous methods. Finally, based on the profound analysis, we highlight the challenges and future trends of deep learning-based PCC, paving the way for further study.

Point cloud media representation format has provided various opportunities for extended reality applications and had become widely used in volumetric content capturing scenarios. At the same time ambiguous storage format representations and network throughput are key problems for wide adoption of this media format. Compression algorithms in corresponding standard activities are aimed to solve this problem. MPEG-I standard has an aim of creating the point cloud compression methodology relying on existing video coding hardware implementations. In scope of the state-of-the-art video-based dynamic point cloud (DPC) compression method, similar 3D patches may be projected in totally different 2D positions in different frames. In this way, the motion vector predictors especially those in the patch boundary may be very inaccurate which may lead to significant bitrate increase. In this paper, we propose to use the reconstructed geometry information to help predict the motion vector more accurately and improve the coding efficiency of the attribute video. First, we propose to use the motion vector of the co-located blocks in the geometry frame as a merge candidate of the current block in the attribute frame. Second, we perform a motion estimation between the current reconstructed point cloud with only the geometry information and the reference point cloud to find the corresponding block. The motion information derived is used as motion vector predictor of the current block in the attribute frame. As far as we can see, this is the first work using the geometry information to compress the attribute in the DPC compression scenario. Significant compression efficiency is achieved with this new 3D point cloud geometry derived motion prediction scheme when compared with the state-of-the-art DPC compression method.

Geometry-based Point Cloud Compression (G-PCC) and Video-based Point Cloud Compression (V-PCC), which are standards developed by MPEG, can be used in compressing the point cloud data. However, these standards do not yet support encoding and decoding in many devices. In order to utilize the point cloud data in many devices, we conducted a study on compression and restoration of the point cloud using legacy codec. Legacy codec has already been used and verified in many devices. And if it is possible to compress and restore point cloud data by utilizing legacy codec, point cloud data can be used on many devices. In this paper, we proposed a method to design point cloud compression based on MP3. We chose MP3 for its proven, popular use, and ubiquitous support of encoder and decoder. We compared proposed method with the G-PCC reference software for performance evaluation.

In this paper, we present a novel 3D structure-awareness image-based point cloud compression scheme, which applies the proposed Symmetry based Convolutional Neural Pyramid (SCNP) to compress colored point clouds view-by-view for 3D model transmission. Input a 3D model to the system, a preprocessing step is first applied to represent the input point cloud as a sequence of view-specific six-dimensional (6D) images, where each pixel is characterized by an RGB color vector and a XYZ 3D point. The transformed 6D images preserve the regular grid structure and thus the redundant information is easy to be removed by conventional image/video compression techniques. Our SCNP first represents each 6D image as a multiple-level pyramid structure for progressively compressing and transmission. The lowest resolution image at the highest level of the pyramid is then decomposed into multiple patches with each of them being coded as the index of a small dictionary through vector quantization. The residual images at other levels are also represented by the vector quantization codes with different patch sizes for progressively reconstructing the input colored point cloud. This process results in a multiple description coding scheme for 3D point cloud compression. With the pre-learned set of dictionaries, the projected view-specific 6D images of the input 3D model are encoded one-by-one to obtain the compressed results for 3D model transmission. In the receiver end, the 3D model is reconstructed by merging all the reconstructed point clouds where each of them is decoded from the corresponding view-specific image. Finally, the conventional 3D reconstruction approach has been applied to remove redundant 3D points for reconstructing the 3D model. Experiments demonstrate the effectiveness of our approach which attains better performance than the current state-of-the-art point cloud compression methods.

Chapter 13 - Point cloud compression

In this article, a survey of the point cloud compression (PCC) methods by organizing them with respect to the data structure, coding representation space, and prediction strategies is presented. Two paramount families of approaches reported in the literature-the projection- and octree-based methods-are proven to be efficient for encoding dense and sparse point clouds, respectively. These approaches are the pillars on which the Moving Picture Experts Group Committee developed two PCC standards published as final international standards in 2020 and early 2021, respectively, under the names: video-based PCC and geometry-based PCC. After surveying the current approaches for PCC, the technologies underlying the two standards are described in detail from an encoder perspective, providing guidance for potential standard implementors. In addition, experiment evaluations in terms of compression performances for both solutions are provided.

Point Cloud Compression (PCC) algorithms can be roughly categorized into: (i) traditional Signal-Processing (SP) based and, more recently, (ii) Machine-Learning (ML) based. PCC algorithms are often evaluated with very different datasets, metrics, and parameters, which in turn makes the evaluation results hard to interpret. In this paper, we propose an open-source benchmark, called PCC Arena, which consists of several point cloud datasets, a suite of performance metrics, and a unified procedure. To demonstrate its practicality, we employ PCC Arena to evaluate three SP-based and one ML-based PCC algorithms. We also conduct a user study to quantify the user experience on rendered objects reconstructed from different PCC algorithms. Several interesting insights are revealed in our evaluations. For example, SP-based PCC algorithms have diverse design objectives and strike different trade-offs between coding efficiency and time complexity. Furthermore, although ML-based PCC algorithms are quite promising, they may suffer from long running time, unscalability to diverse point cloud densities, and high engineering complexity. Nonetheless, ML-based PCC algorithms are worth of more in-depth studies, and PCC Arena will play a critical role in the follow-up research for more interpretable and comparable evaluation results.

Increasing demand for more reliable and safe autonomous driving means that data involved in the various aspects of perception, such as object detection, will become more granular as the number and resolution of sensors progress. Using these data for on-the-fly object detection causes problems related to the computational complexity of onboard processing in autonomous vehicles, leading to a desire to offload computation to roadside infrastructure using vehicle-to-infrastructure communication links. The need to transmit sensor data also arises in the context of vehicle fleets exchanging sensor data, over vehicle-to-vehicle communication links. Some types of sensor data modalities, such as Light Detection and Ranging (LiDAR) point clouds, are so voluminous that their transmission is impractical without data compression. With most emerging autonomous driving implementations being anchored on point cloud data, we propose to evaluate the impact of point cloud compression on object detection. To that end, two different object detection architectures are evaluated using point clouds from the KITTI object dataset: raw point clouds and point clouds compressed with a state-of-the-art encoder and three different compression levels. The analysis is extended to the impact of compression on depth maps generated from images projected from the point clouds, with two conversion methods tested. Results show that low-to-medium levels of compression do not have a major impact on object detection performance, especially for larger objects. Results also show that the impact of point cloud compression is lower when detecting objects using depth maps, placing this particular method of point cloud data representation on a competitive footing compared to raw point cloud data.