基于场景匹配与可重构技术的自适应道路检测

车辆视觉辅助导航系统要求具有很高的鲁棒性和实时性。而视觉处理系统中,单一算法很难克服环境变化产生的影响,做到各种道路环境下的最优。针对目前这一研究难点,提出了基于道路场景匹配和硬件可重构技术的自适应道路检测方法。该方法基于道路场景分类来实现算法与道路场景的最佳匹配,利用可重构硬件来实现硬件算法功能的实时自适应重构,由此实现车辆视觉辅助导航自适应道路检测。实验结果表明:该方法具有检测性能好、鲁棒性强、实时性高的优点,检测正确率达到了90.17%,较好地解决了检测系统自适应性与实时性之间的矛盾。

Self-adaptive road detection based on road scene matching and reconfigurable hardware technique

Vision subsidiary navigation system must be high reality and robustness,but no one algorithm in it can overcome the influence brought by the variational surroundings, and be the best in all kinds of road conditions. The self, adaptive road detection method based on reconfigurable hardware technique and road scene matching is presented, which implements the best matching between algorithms and scenes based on the classification of road scenes,and implements algorithm hardware reconflguration in real time based on reconfigurable hardware, to realizes self-adaptive road detection in vehicle vision navigation. Experiments show it has advantage of good test performance,strong robust,high reality. Detecting precision can reach 90. 17 % ,and it indicated that this method settles the conflict between self-adaptability and reality in detection system.