Abstract:Point clouds offer a novel 3D data representation that has proven pivotal in immersive visual media applications involving human perception. Developing objective point cloud quality assessment (PCQA) methods is imperative, as they can substantially reduce human evaluation costs and drive advancements for visual perceptual experiences in point cloud related applications. Point cloud quality assessment without reference remains challenging. Previous PCQA methods predominantly employ a fixed perceptual distance and often overlook the variability in quality perceived from different viewpoints, which impedes their effectiveness in multiscale or multi-granularity feature extraction and learning, particularly for deep neural networks. The single fixed observation distance fails to capture the multi-resolution characteristics intrinsic to human perception. Addressing this gap, in this paper, we introduce a novel no-reference PCQA method (MOD-PCQA) that integrates multiscale features to enhance point cloud quality perception across diverse scales and granularities. MOD-PCQA pioneers a viewpoint aware feature learning framework, capable of capturing visual features across various granularity levels, from fine to coarse. Specifically, we process and project point clouds into images from different viewpoints. Then, we extract multi-scale features under corresponding perspectives through three branch networks. Finally, we design an alternate learning strategy to optimize the feature extraction network to continuously refine the learned feature information from both inter-scale and intra-scale perspectives. Comprehensive experiments conducted on the SJTU-PCQA and WPC databases validate the superiority of our proposed model over state-of-the-art PCQA methods. Our method achieves optimal performance on both benchmarks by a significant margin, which comprehensively validates its effectiveness for the challenging PCQA task. The source code will be available at https://openi.pcl.ac.cn/OpenPointCloud/MOD-PCQA [1].

Multiscale Spatial Polygonal Object Granularity Factor Matching Method Based on BPNN

Exploring the Synergistic Use of Multi-Scale Image Object Metrics for Land-Use/Land-Cover Mapping Using an Object-Based Approach

Multi-scale Matching Networks for Semantic Correspondence

Complexity-based Matching Between Image Resolution and Map Scale for Multiscale Image-Map Generation.

Multiscale Locality and Rank Preservation for Robust Feature Matching of Remote Sensing Images

A Probabilistic Theory-based Matching Method

A K-Nearest-Neighbor-Pooling Method for Graph Matching

Quasi-dense Matching Based on Multi-features in Continuous Scale for Approximate Epipolar Images

A Feature Matching Method Based on Multi-Level Refinement Strategy

Zoom to Perceive Better: No-reference Point Cloud Quality Assessment via Exploring Effective Multiscale Feature

Multiscale Point Correspondence Using Feature Distribution and Frequency Domain Alignment

Multi-scale Bilateral Aggregation Network for Point Cloud Analysis

Graph-based Robust 3D Point Cloud Map Merging Approach for Large Scale

Spatial Object Recommendation with Hints: When Spatial Granularity Matters

A multiscale road matching method based on hierarchical road meshes

Graph Matching Optimization Network for Point Cloud Registration

A Dual Branch Multiscale Stereo Matching Network for High-Resolution Satellite Remote Sensing Images

A Sub-pixel Mapping Algorithm Based on BP Neural Network with Spatial Autocorrelation Function for Remote Sensing Imagery

Multicore Implementation of the Multi-Scale Adaptive Deep Pyramid Matching Model for Remotely Sensed Image Classification.

A Robust Multisource Remote Sensing Image Matching Method Utilizing Attention and Feature Enhancement Against Noise Interference