基于AdaBoost算法和模板匹配的人眼定位

摘要

以驾驶员疲劳检测为背景,针对人眼的定位问题,提出一种基于AdaBoost算法和模板匹配的人眼定位的方法；该方法首先使用3×3中值滤波去除噪声和光照对图像的影响,然后通过对AdaBoost算法的强分类器训练算法改进、级联分类器优化实现人脸的快速检测,最后在检测到的人脸区域,通过从粗到细改进的模板匹配方法对人眼进行准确定位；实验结果表明,该方法在保证定位准确的情况下,大幅的提高了人眼定位时间;该方法兼具了良好的实时性和鲁棒性.%To driver fatigue detection as the background for the eyes location ,proposes a method of the eyes location based on AdaBoost algorithm and template matching. This method first uses median filtering to remove the impact of image noise and light, then by the strong classifier training algorithm and the cascade optimization of AdaBoost algorithm to achieve rapid detection of human faces. Finally, the detected face region, through improved from coarse to fine template matching method for accurate positioning of the human eyes. The experimental results show that the method to ensure accurate positioning in the case, a substantial improvement of the human eye positioning time. The method combines a good and robust real -time.