基於立體視覺的行人追蹤及計數

Stereo Vision Based People Counting And Tracking

陳至圓(Chih-Yuan Chen)

指導教授 ： 郭忠義

國立臺北科技大學/電資學院/資訊工程系研究所/碩士(2014年)

https://doi.org/10.6841/NTUT.2014.00882

摘要

本論文主要應用於室內及室外之行人計數及追蹤之用途，主要是透過Kinect來進行實作，其原因主要是因為此裝置提供了深度資訊，然而深度資訊幫助了我們突破影像處理技術暨行人計數議題長久以來的許多瓶頸。在本論文中將Kinect架設於量測地點正上方，透過俯瞰式拍攝定位行人的頭頂的移動狀況，並加以分析行人頭部的資訊，進而達成行人的行人計數及追蹤。雖然Kinect提供色彩與深度資訊，本論文參考並實作許多追蹤演算法則，準確度及運算效率仍然不足以即時運算。因此本論文提出行人頭部多重定位演算法則(Multi-Detection Algorithm for Pedestrian Head;MDAPH)，透過深度切片(Depth Slice)、縮圖(Thumbnail)、邊緣偵測(Edge Detection)、霍夫圓形轉換(Hough Circle Transform)、合併頭部資訊(Merge Head Information)、追蹤框演算法(Tracking Window)，成功地進行行人計數。本論文不同於參考文獻的實驗方法，以科學量測方式，更嚴謹地量化行人計數準確度的測量方式，本文提出各種可量化的亮度、情境、異物案例，相較於參考文獻，本論文的實驗結果較為優秀。

關鍵字

行人計數 ； Kinect ； 深度資訊 ； 色彩資訊 ； 俯瞰式拍攝 ； 行人頭頂

並列摘要

This paper uses the Kinect for people counting in indoor/outdoor. Because the Kinect provides color and depth information, we can use it to improve the disadvantages of traditional image processing. We set the Kinect on the top of entrance/exit and use the way of zenithal camera to observe the moving status of pedestrian heads. According to these moving status and the result that is analyzed by color and depth information, we can tracking and counting pedestrians. This proposed method can solve the problem of occlusion and complex scene, and enhance the accuracy of pedestrian analysis. On the other hand, the Kinect uses Infrared ray to detect the distance between the device and objects, so light variance can’t influence our proposed method result.

並列關鍵字

People Counting ； Kinect ； Zenithal Camera ； Color and Depth Information ； Pedestrian heads

參考文獻

[2] A. Destrero, F. Odone, and A. Verri, “A trainable system for face detection in unconstrained environments, ” in Proceedings of the International Conference on Image Analysis and Processing, pp.407-412, 2007.

[3] A. Destrero, F. Odone, and A. Verri, “A system for face detection and tracking in unconstrained environments, ” in Proceedings of IEEE International Conference on Advanced Video and Signal-based Surveillance, pp.499–504, 2007.

[4] B. Antic, D. Letic, D. Culibrk, and V. Crnojevic, “K-means based segmentation for real-time zenithal people counting, ” in Proceedings of International Conference on Image Processing, pp.2565-2568, 2009.

[5] Chao-Ho Chen, Tsong-Yi Chen, Da-Jinn Wang, and Tsang-Kie Chen,“A Cost-Effective People-Counter for a Crowd of Moving People Based on Two-Stage Segmentation, ” Journal of Information Hiding and Multimedia Signal Processing, vol. 3, no. 1, pp. 12-25, 2012.

[7] Che-Hung, Sheng-Luen Chung, and Jing-Ming Guo, “People Tracking in a Building Using Color Histogram Classifiers and Gaussian Weighted Individual Separation Approaches,” Advances in Multimedia Modeling. Springer, pp. 177-186, 2011.