基于特征融合的前方车辆检测

Forward Vehicle Detection Based on Feature Fusion

近年来,由于交通事故发生率逐渐上升,智能交通系统受到研究人员的广泛关注。前方车辆检测作为其中的重要组成部分,能够及时提醒驾驶人员潜在的危险来减少交通事故的发生。基于图像处理技术,针对目前车辆检测方法中鲁棒性差、误检过多的问题,提出一种基于HOG和Haar-like特征融合算法,将提取的特征输入AdaBoost级联分类器进行车辆检测。实验结果表明,本文方法对不同天气情况和道路情况都有很高的准确率和精度且鲁棒性好。

In recent years,due to the increasing traffic accident rate,intelligent traffic system has been widely concerned by researchers.Detection of vehicle in front as one of the important components,can timely remind drivers of potential dangers to reduce the occurrence of traffic accidents.Based on image processing technology,this paper proposes a feature fusion algorithm based on HOG and Haar-like to solve the problem of poor robustness and excessive misdetection in current vehicle detection methods,and input the extracted features into AdaBoost cascade classifier for vehicle detection.Experimental results show that the proposed method has high accuracy and robustness for different weather and road conditions.