论文部分内容阅读
为了提高交通标志分类问题的正确率,提取有效的特征值才可以获得更高的分类正确率。校核分析交通标志图像特点,在分类研究的背景下提出了特征融合的思路,在主成分分析(PCA)降低维度的基础上,提取灰度直方图的特征,将PCA提取的特征和灰度直方图特征融合,并且将融合数据作为分类的输入特征,通过交通标志数据库进行实验分析,多次改变要降低的维度,然后融合灰度直方图特征进行分类,用MATLAB和GUI工具进行仿真,实例验证结果表明,得出的正确率明显提高,在交通标志的分类中效果显著。
In order to improve the correctness of the traffic sign classification problem, we can obtain a higher classification accuracy rate by extracting effective eigenvalues. Based on the analysis of the characteristics of traffic sign images, the idea of feature fusion was put forward in the background of classification research. Based on the principal component analysis (PCA) reduction dimension, the features of gray histogram were extracted, and the features of PCA extraction and gray scale Histogram feature fusion, and the fusion data as the input features of the classification, traffic sign database for experimental analysis, to change the dimensions to be reduced several times, and then fusion gray histogram features classification, using MATLAB and GUI tools for simulation, examples The verification results show that the correct rate is obviously improved and the effect is obvious in the classification of traffic signs.