基于层次轮廓计算机视觉的交通路标识别

Traffic road sign recognition based on computer vision outline level

针对现有计算机视觉对交通路标识别的复杂性和不稳定性的问题,通过运用图像轮廓识别技术,提出了由全局特征到局部特征再到结构特征的多层次轮廓识别,在交通路标的识别过程中,分别构造了图像密度、形状度量、光滑程度和轮廓熵值4个层次的图像轮廓,同时结合Sobel算子和信息熵对交通路标图像进行了提取与分块处理。通过实验仿真结果表明:在图像的提取过程中,交通路标图像随着其DMOS值的增大,图像的质量越差,清晰度越低,其NRSS值越小;在图像的识别过程中,低通滤波器的大小设置为7×7,原图NRSS为0.7654,形状度量为1.3和2.4时,NRSS分别为0.3712和0.2667。这种层次化的轮廓分析在路标的识别上具有较好的稳健性。

For the complexity of problems and instability existing computer vision , traffic signs recognition, through the use of image recognition technology outline presented by the global to the local feature characteristic feature of multi-level structure and then outline recognition , the recognition process of traffic signs the were constructed image density, shape measurement, smoothness and contour entropy four levels of image contours, combined with Sobel operator and traffic signs image information entropy were extracted and processed block. The simulation results show that：in the extraction process of an image, the image with the increasing traffic signs DMOS its value, the worse the image quality, the lower the resolution, the smaller the NRSS value;in the recognition process of the image, the low when the size of the set-pass filter is 7 ＆#215; 7, picture NRSS to 0.7654, shape of metric is 1.3 and 2.4, NRSS 0.3712 and 0.2667 respectively. This level of profiling has better robustness in recognition of road signs.