EGMS-PCQM: Entropy-Guided Multiscale Sampling-Based Dual-Net for No-Reference Point Cloud Quality Measurement
IEEE Transactions on Instrumentation and Measurement · 2026 · 75 · 1-17
DOI: 10.1109/tim.2026.3667351 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Instrumentation and Measurement, 2026. Verfasst von Shenglong Liu und 6 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 67 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Zhejiang Provincial Natural Science Foundation of China, Ningbo Innovation “Yongjiang 2035” Key Research and Development Program.
Abstract (Originalsprache)
Point clouds (PCs) are vital for accurate 3-D modeling of the real world, yet distortions during transmission and generation impact their quality. Developing no-reference PC quality measurement (NR-PCQM) methods is essential for determining whether PC’s visual quality meets the application requirements. Most existing NR-PCQM methods rely on sampling strategies like farthest point sampling (FPS) or random sampling (RS), which inadequately simulate the human visual system (HVS), particularly its sensitivity to details, textures, and multiscale perception. To address these challenges, we propose a multiscale NR-PCQM framework named entropy-guided multiscale sampling (EGMS)-PCQM, which employs EGMS to prioritize clusters rich in complex textures relevant to HVS and incorporates both small-scale clusters to capture local details and large-scale clusters to encompass broader contextual information. Then, a dual-net framework, consisting of the small-scale feature extraction network (SSFE-Net) and large-scale feature extraction network (LSFE-Net), is proposed to process the regions of interest identified by the HVS through the sampling process. SSFE-Net extracts fine-grained features from localized regions using dual local maps and a progressive strategy, while LSFE-Net extracts coarse-grained features from large-scale clusters by decoupling edge and content regions with an adaptive attention mechanism aligned to HVS sensitivities. The experimental results demonstrate that the proposed method achieves superior performance, with SRCC gains of 0.75% and 4.28% over the second-best method on the SJTU-PCQA and WPC datasets, respectively. The source code will be available at https://github.com/ll2s/EGMS-PCQM
Förderung
- National Natural Science Foundation of China
- Zhejiang Provincial Natural Science Foundation of China
- Ningbo Innovation “Yongjiang 2035” Key Research and Development Program
- Natural Science Foundation of Ningbo Municipality
- Laboratory of Intelligent Home Appliances, College of Science and Technology, Ningbo University
- Ningbo Public Welfare Research Project
Zitieren
APA
Liu, S., He, Z., Jiang, G., Jin, C., Zhou, W., Zhu, L., & Luo, T. (2026). EGMS-PCQM: Entropy-Guided Multiscale Sampling-Based Dual-Net for No-Reference Point Cloud Quality Measurement. IEEE Transactions on Instrumentation and Measurement, 75, 1-17. https://doi.org/10.1109/tim.2026.3667351
Vancouver
Liu S, He Z, Jiang G, Jin C, Zhou W, Zhu L, et al. EGMS-PCQM: Entropy-Guided Multiscale Sampling-Based Dual-Net for No-Reference Point Cloud Quality Measurement. IEEE Transactions on Instrumentation and Measurement. 2026;75:1-17. doi:10.1109/tim.2026.3667351
BibTeX
@article{Liu2026,
author = {Liu, Shenglong and He, Zhouyan and Jiang, Gangyi and Jin, Chongchong and Zhou, Wujie and Zhu, Linwei and Luo, Ting},
title = {EGMS-PCQM: Entropy-Guided Multiscale Sampling-Based Dual-Net for No-Reference Point Cloud Quality Measurement},
journal = {IEEE Transactions on Instrumentation and Measurement},
year = {2026},
volume = {75},
pages = {1--17},
doi = {10.1109/tim.2026.3667351}
}
EUVIMED