以FPGA實現即時人臉追蹤系統

FPGA Implementation of a Real-Time Face Tracking System

黎陳秀(Tran-Tu Le)

指導教授 ： 徐元寶

國立虎尾科技大學/電機資訊學院/資訊工程研究所/碩士(2013年)

https://doi.org/10.6827/NFU.2013.00056

摘要

本文提出了一種基於FPGA平台的即時人臉追蹤系统。在系统中，人臉的偵測主要是基於臉部的基本特徵：膚色。首先，圖像是由CMOS攝影機捕獲，接著經由子樣本模塊來減少圖像的複雜性。其次，將RGB格式的輸入圖像變換到對數色系座標系統(log-opponent)的顏色空間，並經由二進制膚色概率值的選定來過濾膚色像素。第三，本系統利用關連分量分析演算法(connected component analysis algorithm)來處理多皮膚區域的情況。第四，皮膚區域被順序地檢查以確定是否它們是人類臉部。最後，以綠色的方形框框住被偵測出的人臉以鎖定人臉，並在VGA螢幕上顯示。同時，控制安裝有CMOS攝影機的基座上的馬達以追蹤鎖定的人臉。實驗結果顯示，本系统具有良好的性能和满足人臉追蹤的目標。

關鍵字

FPGA ； 人臉追蹤 ； log-opponent

並列摘要

This thesis presents a FPGA based platform for a real-time face detection and tracking system. In the system, human faces can be detected based on a particular facial feature, the skin color. Firstly, images are captured by a CMOS camera and a sub-sample module is used to reduce the images complexity. Secondly, the RGB format input image is transformed to log-opponent color space. A binary skin probability value is chosen to filter the skin pixels. Thirdly, a connected component analysis algorithm is applied to separate skin regions if the image has more than one skin regions. Fourthly, the skin regions are sequentially checked to determine if they are human faces or not. Finally, face tracking windows will be built to lock the human faces and displayed on the VGA screen. Meanwhile, a surveillance machinery is controlled, on which the CMOS camera is mounted, for tracking the prominent face in multi-angle. Experimental results show that the proposed system achieves high performance and satisfies the goal of real-time face tracking.

並列關鍵字

FPGA ； face tracking ； log-opponen

參考文獻

[23] C. Y. Chen, H. C. Huang, R. C. Hwang, “A low complexity real time face tracking system with fuzzy controller,” International Journal of System, Volume 8, Issue 4, December 2006.

[1] S. Paschalakis, M. Bober, “Real-time face detection and tracking for mobile videoconferencing,” Real-Time Imaging, Volume 10, Issue 2, Pages 81-94, April 2004.

[2] E. Hjelmas, B. K. Low, “Face detection: a survey,” Computer Vision and Image Understanding, Volume 83, Issue 3, Pages 236-274, September 2001.

[4] M. Turk, A. Pentland, “Eigenfaces for recognition”, Journal of Cognitive Neuroscience, Volume 3, Issue 1, Pages 71-86, 1991.

[5] H. A. Rowley, S. Baluja, T. Kanade, “Neural-network based face detection,” IEEE Transactions, PAMI, January 1998.