一种基于灰关联和Sobel算子的完全非结构化道路边缘检测方法

Completely Unstructured Road Edge Detection Based on Grey Relation and Soble Operator

完全非结构化道路检测是智能车辆自主行驶所面临的关键技术难题.图像的边缘是图像特征提取与分析理解的基础,其检测质量直接决定后期理解效果.基于灰关联理论和Sobel算子,针对实时道路图像的真彩色信息,提出一种有效的边缘提取策略.仿真结果表明,直接利用Sobel算子对彩色图像进行边缘检测,不但图像边缘比较清晰,而且还能有效降低图像的噪声.

Solving completely unstructured road detection problem may improve the adaptability for environment,which is the key for intelligent vehicle driving. Image edge is the foundation of feature extraction, analysis and comprehension. Its quality of image determineds the processing results. Conventional edge detection algorithms suffer from the shortcomings of undetected and artificial information. Based on the theory of grey relation analysis and Sobel operator, the paper proposes a edge detection algorithm for the true-color information of real-time road image. The simulation results show that image edge is more ambiguous if Sobel operator for color image edge detection is directly used. They also show that the image edge is relatively clear and can effectively reduce the noise of the image.