论文部分内容阅读
介绍一种基于Gabor特征和多分辨率的车辆检测方法,该方法首先在假设产生阶段根据道路场景图像的消失点确定图像的兴趣区域,以垂直和水平边缘为依据产生相应兴趣区域的假设链,最后将各兴趣区域假设链合并,产生最终的假设,验证阶段用支撑向量机分类器验证假设正确与否,在保证鲁棒性的同时,提高实时性,此方法在假设产生阶段大大减少非兴趣区域对系统计算资源的消耗,减少计算负担,且在假设验证阶段有效减少伪目标对检测率的影响。实验表明,本文算法处理速度可达20帧/s,检测率在90%以上。
This paper introduces a vehicle detection method based on Gabor features and multi-resolution. Firstly, the region of interest of the image is determined according to the vanishing point of the road scene image at the stage of hypothesis generation, and the hypothesis chain of the corresponding region of interest is generated based on the vertical and horizontal edges. Finally, the hypothesis is merged with the hypothesis chains in each region of interest, and the support vector machine classifier is used to verify the hypothesis in the verification stage. While assuring the robustness, real-time performance is improved. This method greatly reduces the non-interest in the hypothesis generation phase The region consumes the system computing resources, reduces the computational load, and effectively reduces the impact of false targets on the detection rate in the hypothesis verification stage. Experiments show that this algorithm can process up to 20 frames / s, and the detection rate is over 90%.