Clustering and DCT Based Color Point Cloud Compression

详细信息    查看全文

• 作者：Ximin Zhang ; Wanggen Wan ; Xuandong An
• 关键词：Color point cloud ; Mean shift cluster ; Project ; DCT transform
• 刊名：Journal of Signal Processing Systems
• 出版年：2017
• 出版时间：January 2017
• 年：2017
• 卷：86
• 期：1
• 页码：41-49
• 全文大小：
• 刊物类别：Engineering
• 刊物主题：Signal,Image and Speech Processing; Circuits and Systems; Electrical Engineering; Image Processing and Computer Vision; Pattern Recognition; Computer Imaging, Vision, Pattern Recognition and Graphics;
• 出版者：Springer US
• ISSN：1939-8115
• 卷排序：86

文摘

In this paper, a new point cloud compression method is proposed. The 3D color point cloud is firstly mean-shift clustered into many homogeneous blocks based on the similar spatial (XYZ) information of each point. Based on the RANdom SAmple Consensus (RANSAC) algorithm, those points being clustered in the same block are fitted by a 3D plane and all these points belonging to the same block are projected to this corresponding plane. For every plane an optimal rectangle bounding box is identified and is divided into n × n grids, the color (RGB) information associated with each grid point is replaced by the average of RGB values of all the projected points falling in this grid. Finally, a 2D DCT (Discrete Cosine Transform) transform is performed on these n × n grids points. The compressing ratio can reach 32 with negligible spatial and color distortion.