基于方向分形维的高速公路车辆边缘检测研究

Research for Vehicle Edge Detection on Highway Based on Direction-Fractal Dimension

基于图像的高速公路障碍物检测是实现汽车视觉导航的核心问题。在一个景物快速变化的视频图像中检测出目标物体是难度较大的图像分割和识别问题,而目前的各种分割算法也各有优劣。本文主要针对四川和重庆地区高速公路的特点,在分形维的基础上提出了方向分形维的算法,结合图像的灰度特征,用来提取车辆边缘并分割高速公路上的车辆。处理结果显示了方向分形维较强的分割能力和对特殊背景噪声的鲁棒性。本文针对分形算法的计算特点构造了相应的快速算法。经上路实验验证,算法实时性满足高速公路视觉导航系统的需要。

Obstacle detection is the key problem in vehicle vision navigation system on highway. To segment and recognize the object in changeable background is a very difficult problem. Based on the fractal dimension, this paper describes the direction-fractal dimension algorithm to segment the vehicle on highway in Sichuan province and Chongqing city. We give the material algorithm and quick algorithm about the direction-fractal dimension in this paper. The result shows that fractal and direction-fractal dimension has strong robust and strong ability of segmentation. Based on fractal dimension, the vision navigation system could detect the vehicle perfectly.