基于机器视觉的非结构化道路检测算法研究  

Research on unstructured road detection algorithm based on machine vision

作　　者：魏武[1] 龚树锋[1] 

机构地区：[1]华南理工大学自动化科学与工程学院,广东广州510640

出　　处：《传感器与微系统》2010年第1期39-41,共3页Transducer and Microsystem Technologies

摘　　要：提出了一种提高非结构化道路边缘检测实时性和准确性的算法。该算法中首先对原始道路图像进行中值滤波,抑制随机噪声,并选择基于双峰法的多阈值Otsu方法进行图像分割,使分割效果和分割时间得到优化。通过Canny算子进行初次边缘检测,然后,采用数学形态学修正,从而得到完整、清晰的道路边缘图像。仿真结果表明:该非结构化道路检测方法具有良好的实时性和准确性。A kind of unstructured road detection algorithm which has improved real-time detection and accuracy is introduced. First, the original images are processed by median-filtering, and random noise is suppressed, and an Otsu multi-threshold algorithm is chosen based on two-peak method for rapid image segmentation, the segmentation effect and the segmentation time are optimized;Finally the primary edge detection is carried on with the Canny operator,and mathematics morphology is used to revise,thus obtain the complete and clear path edge image. The simulation result indicates that this unstructured path examination method has good characteristics of timeless and accuracy.

关 键 词：移动机器人 机器视觉 非结构化道路 边缘检测 

分 类 号：TP317.4[自动化与计算机技术—计算机软件与理论]