Abstract:

Traffic sign recognition plays an important role in safe driving, especially under the condition of illumination change and occlusion. Automatic recognition of traffic signs with high accuracy and high real-time performance needs to be resolved. This paper reviews the detection of traffic sign and the recognition of traffic sign. And the principles, steps, characteristics and performances of the algorithms are introduced, as well as the common traffic sign databases for algorithm research. Compared with traditional detection and recognition methods, deep learning algorithms is helpful to solve the difficult problems of traffic sign recognition under the condition of illumination change and partial occlusion, which is the main way to improve the performance of automatic driving and unmanned driving in the future.

Key words: traffic sign detection, traffic sign recognition, deep learning

摘要：

交通标志识别对于车辆安全行驶具有重要作用，特别是针对存在光照变化和遮挡的情况，准确性高、实时性好的交通标志自动识别亟需解决。对交通标志的检测和交通标志的识别分别进行了综述，给出了它们的原理、步骤、特点和性能，以及进行算法研究常用的交通标志数据库。相比于传统的特征检测和识别方法，采用深度学习有助于解决光照变化、部分遮挡等情况下的交通标志识别难题，是今后自动驾驶和无人驾驶性能提升的主要途径。

关键词: 交通标志检测, 交通标志识别, 深度学习