基於 3D LiDAR 之室内人體定位與辨識

Indoor Human Localization and Recognition Using 3D LiDAR

林經緯(WELLY SETIAWAN LIMANTORO)

指導教授 ： 曾煜棋

國立交通大學/電機學院/電機資訊國際學位學程/碩士(2019年)

摘要

室內的人體定位與辨識是電腦視覺領域的一項重要研究課題。其中，最常採用的設備是較為經濟實惠的攝影機。然而，攝影機容易受到光線的影響, 如: 在低光源的環境不易辨識畫面資訊。我們發現光學雷達(LiDAR)會主動發出光線進行測量, 更適用於不同光線條件下的物件辨識測與定位任務。
在本論文中, 我們使用 3D LiDAR 來建立一套在室內環境下的人體定位識別系統。同時，我們選用具有 16個通道的 3D LiDAR(型號: Velodyne VLP-16)開展了大量的實驗量測。本論文的主要貢獻包括:（1）使用一台低端的 3D LiDAR 實現了室內環境下的人體定位與識別；（2）收集了不同情境下的的點雲資訊,並建立了包含不同姿勢的人體外型與非人體的物件資訊的數據集；（3）建立了雛型系統，並通過大量實驗來驗證不同場景下所提方法的穩健性。從實驗結果得知, 本系統的識別正確性高達98.18%。

關鍵字

光達 ； 人體定位與辨識

並列摘要

Indoor human localization and recognition are fundamental issues in computer vision. Cameras are most commonly used to solve these issues at an affordable price. However, a camera has its limitations, especially in dark conditions, because it relies on surrounding lights to capture data. Light Detection and Ranging (LiDAR) devices, on the contrary, are based on self-emitted light, and they are more suitable for performing detection and ranging tasks under various light situations.
In this work, we propose a human localization and recognition system in an indoor environment by using a 3D LiDAR device. We conduct a number of experiments by using 3D LiDAR (Velodyne VLP-16) which has 16 channels. The main contributions of this thesis work include (i) proposing a system for human localization and recognition in an indoor area by using a low-end 3D LiDAR, (ii) collecting a 3D point dataset of human objects in various poses as well as non-human objects in various scenarios, and (iii) implementing a prototype to verify the robustness of our approach. The accuracy of human localization and recognition system is 98.18%.

並列關鍵字

LiDAR ； Human Localization and Recognition

參考文獻

[1] K. Kobayashi, K. Iwamura, K. Kaneda and I. Echizen, "Surveillance Camera System to Achieve Privacy Protection and Crime Prevention," in Proceedings of International Conference on Intelligent Information Hiding and Multimedia Signal Processing, 2014. pp. 463-466.

[2] C. R. Gonzalez, A. Villena, D. G. Medina, J. M. Gomez, L. Rodriguez and I. G. Varea, "InLiDA: A 3D Dataset for People Detection and Tracking in Indoor Environments," in Proocedings of International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications, 2017. pp. 484-491.

[3] Z. Yan, T. Duckett and N. Bellotto, "Online Learning for Human Classification in 3D LiDAR-based Tracking," in Proceedings of International Conference on Intelligent Robots and Systems, 2017. pp. 864-871.

[4] J. Shackleton, B. VanVoorst and J. Hesch, "Tracking People with a 360-Degree Lidar," in Proceedings of International Conference on Advanced Video and Signal Based Surveillance, 2010. pp. 420-426.

[5] K. Kidono, T. Miyasaka, A. Watanabe, T. Naito and J. Miura, "Pedestrian Recognition Using High-definition LIDAR," in Proceedings of Intelligent Vehicles Symposium, 2011. pp. 405-410.