一個以圖形識別為基礎的即時視線追蹤及眨眼辨識系統與其在嵌入式系統之實現

A real-time eye tracking and blinking detection system based on pattern recognition and its implementation on embedded system

王嘉誠(Jia-Cheng Wang)

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(2012年)

https://doi.org/10.6841/NTUT.2012.00515

摘要

隨著電腦設備的不斷進步，機器圖形識別的準確率與速度都不斷提升，近年來許多學者也不斷投入這方面的研究。由於現今嵌入式系統的發展，網路攝影機和擁有照相機功能手機已相當普及，且效能已足夠滿足即時辨識系統的運算量。因此，我們提出一套即時人眼辨識與追蹤系統及人機互動機制，可以運用在協助老人或是提供如漸凍人等無法自由使用滑鼠鍵盤的人，另一種操作電腦的方式。
在系統運作的過程中，可能會遇到光線明暗、環境背景、拍攝的角度、影像雜訊等問題，本論文運用了影像強化(加強辨識率)、Retinex光線處理(減弱光線的影響)、AdaBoost演算法結合Haar-like features的圖形辨識(人臉與人眼辨識技術)等來完成本套系統，並在嵌入式開發板(OMAP3530)上實現。實驗結果顯示，本論文所提之系統架構能夠有效且即時的追蹤出眼球位置，並能偵測眨眼的狀態，進而做出人機互動。即使在光線不均勻的情況下，系統也能達到成功且高正確率的辨識效果。

關鍵字

AdaBoost ； Retinex ； 嵌入式系統 ； 圖形識別 ； 人眼辨識 ； 人眼追蹤

並列摘要

Due to the progress of computer technology, the detecting accuracy and speed of pattern recognition are promoted. Additionally, according to research advances in embedded systems, the applications of digital cameras, such as internet cameras and smart phones, are widely used in nowadays. Therefore, we propose a set of real-time human-eye recognition and tracking systems with human-computer interaction mechanism. It can be adopted to help the users with Amyotrophic Lateral Sclerosis (ALS), elderly individuals, and other people who cannot use the mouse or keyboard freely with another way to access the computers.
So far, there still exist some difficulties in practical usage such as light shades, environmental background, camera angles, image noise and other issues. To overcome these problems, this study employees image enhancement, the Retinex method for light processing, and AdaBoost algorithm combining with Haar-like features to enhance the detection accuracy, to reduce the influence of light, and to achieve human-face and human–eye recognition. The proposed system is demonstrated on OMAP3530 for embedded applications, and experimental results show that the proposed architecture is capable of effective and real-time eye position tracking. The proposed system can also detect the status of eye blink. Even for the case of uneven lighting, the proposed system can successfully recognize the human eyes with high accuracy.

並列關鍵字

AdaBoost ； Retinex ； embedded system ； Pattern Recognition ； Eye Recognition ； Eye Tracking

參考文獻

[26] 陳庚延，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，碩士論文，國立台北科技大學，台北，2011。

[27] 何治宇，應用連通元件標記法於即時人眼追蹤，碩士論文，國立台北科技大學，台北，2010。

[4] R. M. Haralick, S. Sternberg, and X. Zhuang, “Image analysis using mathematical morphology,” IEEE Pattern Anal. Machine Intell., vol. PAMI-9, pp. 532–550, July 1987.

[6] N. Dalal and B. Triggs, “Histograms of oriented gradients for human detection,”Computer Vision and Pattern Recognition, USA, pp.20-25, 2005.

[7] H. X. Jia and Y. J. Zhang, “Fast human detection by boosting histograms of oriented gradients,” IEEE Image and Graphics, pp.683-688, 2007.