基於距離影像的光達點雲幀間壓縮

Range Image-based Inter-frame Compression for LiDAR Point Clouds

劉佳昇(Chia-Sheng Liu)

指導教授 ： 李明穗徐宏民

國立臺灣大學/電機資訊學院/資訊工程學系/碩士(2022年)

https://doi.org/10.6342/NTU202204094

摘要

光達傳感器所收集的大量資料也隨之產生了光達資料壓縮的問題。而基於距離影像的光達資料壓縮方法是一個有潛力的候選解決方案，但是在幀間編碼方面缺乏充分的探索。在我們的工作中，我們嘗試處理基於距離影像的光達傳感器資料幀間壓縮問題。我們的幀間壓縮框架遵循預測編碼的典範，該典範已在彩色影像編碼領域被採用。我們提出的框架包括一個預測模組和一個殘差編碼模組。對於預測模組，我們將時間序列中的前一幀和後一幀作為參考幀，並對將要編碼的幀進行幀級預測。至於殘差編碼模組，與以往先量化再進行無損編碼的方法不同，我們引入了一種基於學習的方法，可以進一步開發殘差幀內的空間冗餘性。在SemanticKITTI資料集上進行的實驗顯示出，在低碼率的條件之下，我們的方法優於其他基於距離影像的方法。

關鍵字

壓縮 ； 光達 ； 點雲 ； 距離影像

並列摘要

The large amount of data collected by LiDAR sensors brings the issue of LiDAR data compression. Range image-based method for LiDAR data compression is a potential candidate for solution but lacks full exploration, especially for inter-frame coding. In our work, we address the problem of range image-based LiDAR sensor data inter-frame compression. Our inter-frame compression framework follows the predictive coding paradigm, which is adopted in the field of color video codec. Our proposed framework consists of a prediction module and a residual coding module. For the prediction module, we take the previous and next frame in time steps as reference frames, aiming to perform a frame-level prediction of the to-be-encoded frame. As for the residual coding module, different from previous methods that first apply quantization followed by a lossless coder, we introduce a learning-based approach that can further exploit spatial redundancy within the residual frame. Experiments conducted on the SemanticKITTI dataset demonstrate that our method outperforms other range image-based methods, especially at low bitrate conditions.

並列關鍵字

Compression ； LiDAR ； Point Clouds ； Range Image

參考文獻

E. Agustsson, D. Minnen, N. Johnston, J. Balle, S. J. Hwang, and G. Toderici. Scale-space flow for end-to-end optimized video compression. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020.

J. Ballé, D. Minnen, S. Singh, S. J. Hwang, and N. Johnston. Variational image compression with a scale hyperprior. In International Conference on Learning Representations (ICLR), 2018.

J. Behley, M. Garbade, A. Milioto, J. Quenzel, S. Behnke, C. Stachniss, and J. Gall. SemanticKITTI: A Dataset for Semantic Scene Understanding of LiDAR Sequences. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2019.

S.Biswas, J.Liu, K.Wong, S.Wang, and R.Urtasun. Muscle: Multi sweep compression of lidar using deep entropy models. Advances in Neural Information Processing Systems, 2020.

Z. Cheng, H. Sun, M. Takeuchi, and J. Katto. Learned image compression with discretized gaussian mixture likelihoods and attention modules. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020.