The horizontal features of vehicle bottom and the vertical features of both sides of vehicle are selected, and the local features of gray scale and its gradient and variation, which are less affected by global gray scale, are extracted. The horizontal and vertical features of vehicle are fused respectively by applying weighted evidence theory, and the weight of each feature is adjusted according to different light intensity. For improving the real-timeness of identification, artificial fish swarm algorithm is applied, and a vehicle identification module is added to enhance the guiding ability of artificial fishes. Based on search results and further judgment by symmetric feature, the accurate positioning of front vehicle is realized. Experiment results show that the technique proposed can accurately identify front vehicles under different lighting conditions with good adaptability, accuracy and realtimeness.%选取车辆底部水平方向特征和车辆左右两侧垂直方向特征,提取出受全局灰度影响较小的局部灰度特征、局部梯度特征和局部波动特征.应用加权证据理论将车辆水平和垂直方向的特征分别进行信息融合,并根据光强的不同调整各特征的权重.为提高识别的实时性,将人工鱼群算法应用于识别中,并增添了车辆识别模块以增强人工鱼的引导能力.根据人工鱼群的搜索结果,再通过对称性特征进行进一步判定,实现前方车辆的准确定位.实验结果表明本方法可在不同光照条件下对前方车辆进行识别并准确定位,具有良好的适应性、准确性和实时性.
展开▼
