3D Color Point Cloud Compression with Plane fitting and Discrete Wavelet Transform

Abstract

Significant advances of 3D data sensors such as 3D lasers and the Microsoft Kinect provide a massive data which is yielding synchronized depth and color information. It is necessary to develop the new methods to reduce the time, bandwidth and increase the memory space for transmitting the dense point cloud. This proposed work has been concentrated to reduce the spatial information and color information. The spatial information has been reduced by using Adaptive Planar Outlier Removal (APOR) function and the color information has been compressed based on Haar Discrete Wavelet Transform (DWT). A novel quantization method of this proposed work is planar outlier removal function on spatial information compression and unit vector normalization to normalize the 3D color vector coefficients. Experimental results show that the proposed work performed well on the Cr component of the sample images with high PSNR, less compression time and preserved the entropy value of the original point cloud. Proposed methodology compared with the existing Discrete Cosine Transform (DCT) based compression. The proposed methodology produced the PSNR value approximately 10 times better than existing DCT compression technique.