一個以視覺辨識為基礎之人眼即時偵測系統與其在嵌入式系統之實現

A real-time eyes detection system based on visual identification and its implementation on embedded systems

陳庚延(Geng-Yen Chen)

指導教授 ： 陳彥霖

國立臺北科技大學/電資學院/資訊工程系研究所/碩士(2011年)

https://doi.org/10.6841/NTUT.2011.00432

摘要

近年來在利用電腦視覺技術於人眼偵測上之研究，已有相當多的學者投入開發，但是實際的偵測系統是必須運作在複雜的場景中，為了達到系統的即時性，環境中複雜的背景、不均勻的亮度、整體環境照明度、不同的拍攝角度等，都會嚴重的影響到人臉及人眼的偵測率及偵測時間。因此本論文之目的在於提出一個以低成本方式同時可抵抗照度不均勻的情況下實作一套人眼即時偵測系統，並在嵌入式平台上實現。此系統大致上可分成以下五種技術：影像增強技術、不均勻照度還原技術、人臉偵測技術、眼睛偵測技術。其中影像增強部分是使用直方圖等化來增強對比度，不均勻照度是使用Retinex演算法來降低光線所造成的影響，在人臉及人眼之偵測是透過Haar-like features結合AdaBoost learning algorithm之特徵辨識方法，該方法可有效抵抗不同膚色人種之問題。
實驗結果顯示，本論文所提出來的架構，能克服光線不均勻之問題，偵測準確率可達90%，偏差量在3cm內，然而在偵測速度上依距離webcam之遠近的不同可達每秒10張~20張畫面。

關鍵字

電腦視覺 ； 影像增強 ； Retinex ； AdaBoost ； 嵌入式系統

並列摘要

Recently, there are many studies focus on the eye detection based on the computer vision technology. However, the eye detection system in the practical usage should work in the complex environment robustly and correctly. Besides, the efficiency of the eye detection system is the critical issue in order to achieve the system in real time. The eye detection system must overcome the complex background, the uneven brightness, the overall environmental illumination, and the different camera angles problems. These conditions will seriously affect the detection rate and detection time.
Therefore, we present a low cost real-time eye detection which conquers the non-uniform and complex environmental illumination. The proposed framework can be divided into the following five technologies: the image enhancement technology, the reduction of the impact of uneven illumination technology, the face detection technology, and the eye detection technology. We use the Retinex algorithm to eliminate the impact of light, and then use the Haar-like features with AdaBoost learning algorithm to achieve face and eye detection. The proposed method is also implemented on the embedded system in real time. The experimental results demonstrate that the detection accuracy is over 90% with 10 ~ 20fps frame rate.

並列關鍵字

Computer vision ； image enhancement ； Retinex ； AdaBoost ； embedded systems

參考文獻

[4] H. T. Wang, S. Z. Li and Y. S. Wang, "Face recognition under varying lighting conditions using self quotient image," Proceedings of the Sixth IEEE International Conference on Automatic Face and Gesture Recognition Proceedings, 2004, pp.17-19.

[5] P. N. Belhumeur and D. J. Kriegman, "What is the set of images of an object under all possible lighting conditions?," Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, 1997, pp. 52-58.

[6] M. Lee and C. H. Park, "An efficient image normalization Method for face recognition under varying illuminations," IEEE Transactions on Pattern Analysis and Machine Intelligence, 1997, pp.711-720.

[8] M. H. Yang, "Detection Faces In Image : A Survey, Pattern analysis and Machine," IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 24, 2002, pp. 34-58.

[9] G. Yang and T. S. Huang, "Human Face Detection in Complex Background," Pattern Recognition, Vol. 27, No. 1, 1994, pp. 53-63.