论文部分内容阅读
交通标志识别系统（Traffic Sign Recognition）是智能交通系统（IntelligentTransport Systems）的重要研究方向。道路交通环境的复杂性和交通标志数据库规模庞大等原因使得在设计交通标志识别系统可行性方案时必须考虑到时间复杂度和识别率。本文提出了一种高效且快速的基于改进主成份分析法（PCA-principal component analysis ）和极限学习机（ELM -extreme learning machine）的交通标志识别算法，简称为 PCA-ELM。该算法首先提取出交通标志数据库中每个交通标志的梯度方向直方图（HOG-gradient direction of histogram）特征，然后利用改进 PCA 算法对提取出的 HOG 特征进行降维处理。之后利用降维后的 HOG 特征进行 ELM 模型训练，最后利用经过训练的 ELM 模型对测试图片进行识别。实验结果表明，基于PCA-HOG 和 ELM 模型的交通标志识别算法在获得低时间复杂度的同时可以得到 97.69%的识别率。
Traffic Sign Recognition (Traffic Sign Recognition) is an important research direction of Intelligent Transportation Systems. The complexity of road traffic environment and the large size of traffic sign database make time complexity and recognition rate must be taken into consideration when designing traffic sign recognition system feasibility plan. This paper presents an efficient and rapid traffic sign recognition algorithm based on PCA-ELM and ELM-extrem learning machine (PCA-ELM). Firstly, the HOG-gradient direction of histogram feature of each traffic sign in the traffic sign database is extracted. Then, the improved PCA algorithm is used to reduce the dimensionality of the extracted HOG features. After that, the ELM model was trained using the HOG features after dimensionality reduction. Finally, the trained ELM model was used to identify the test images. The experimental results show that the traffic sign recognition algorithm based on PCA-HOG and ELM model can achieve 97.69% recognition rate while obtaining low time complexity.