LS-PCQA
收藏资源简介:
LS-PCQA数据集由上海交通大学合作媒体创新中心创建,包含104个高质量参考点云和超过22,568个失真样本。每个参考点云通过31种不同类型的失真(如高斯噪声、对比度失真、局部缺失和压缩损失)在7个失真级别上进行处理。数据集旨在支持无参考点云质量评估(NR-PCQA)的研究,特别是在点云的生成、压缩、传输和展示过程中可能出现的各种失真。通过大规模的主观实验和伪MOS评分,数据集为开发和评估新的NR-PCQA算法提供了丰富的资源。
The LS-PCQA dataset was created by the Collaborative Media Innovation Center of Shanghai Jiao Tong University. It comprises 104 high-quality reference point clouds and more than 22,568 distorted samples. Each reference point cloud is subjected to distortions of 31 distinct types across 7 distortion levels, including Gaussian noise, contrast distortion, local missing, and compression loss. This dataset is intended to support research on no-reference point cloud quality assessment (NR-PCQA), particularly for various distortions that may occur during the generation, compression, transmission, and display of point clouds. Through large-scale subjective experiments and pseudo-MOS scoring, the dataset provides a rich resource for developing and evaluating new NR-PCQA algorithms.




