Lidar Range Image Compression Using Deep Delta Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data compression techniques for lidar sensors focus on compressing generic 3D point clouds, failing to leverage the lidar scanning pattern, leading to inefficient storage and transmission costs in applications like autonomous driving and augmented reality.
Innovation Solution
Directly compressing lidar range images using quantization, deep delta encoding, and entropy encoding, leveraging the lidar scanning pattern to improve compression quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If generic 3D point cloud compression techniques are used, then compression can be applied to lidar data, but compression quality is suboptimal because the lidar scanning pattern is not leveraged
Solution Approach 1:
The patent segments the lidar range image into multiple blocks and processes each block independently through the compression pipeline (quantization, deep delta encoding, entropy encoding). This segmentation allows the algorithm to exploit local scanning patterns while maintaining computational efficiency, resolving the contradiction between achieving high compression quality and managing algorithmic complexity.
Solution Approach 2:
The patent applies preliminary quantization to the range image values before deep delta encoding. This preliminary action reduces the dynamic range of the data early in the pipeline, making subsequent compression steps more effective. By performing this preparation step beforehand, the algorithm achieves better compression quality without proportionally increasing overall complexity.
2Quantity of substance
If compression is applied to reduce data volume, then storage and transmission costs decrease, but compression algorithms must process large datasets efficiently
Solution Approach 1:
By dividing the large lidar range image into smaller blocks that can be processed independently and in parallel, the patent maintains high compression ratios while improving processing throughput. This segmentation enables efficient utilization of computational resources, resolving the contradiction between reducing data volume and maintaining compression productivity.
Solution Approach 2:
The patent applies aggressive quantization that discards certain fine-grained details in exchange for significantly reduced data volume. This partial action approach achieves substantial compression ratios while keeping the processing complexity manageable, effectively balancing data volume reduction with processing efficiency.
3Loss of information
If deep delta encoding with multiple previous blocks is used, then compression quality improves, but computational complexity increases
Solution Approach 1:
The patent uses deep delta encoding that references a limited number of previous blocks (e.g., 2-4 blocks) rather than the entire history of compressed data. This partial action approach captures the essential temporal correlations in lidar scans achieved by leveraging the scanning pattern, while constraining the computational energy required to a manageable level by not examining all previous blocks.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for performing data compression and data decompression on lidar range images.