摘要
介绍一种在MATLAB环境下彩色立体视觉装置在环的实时图像处理系统 ,用于自动识别前方汽车目标和计算距离。鉴于汽车均具有红色尾灯这一普遍特征和人观测色彩的一致性 ,在此采用了基于色度—饱和度—亮度模型的模糊规则来提取汽车尾灯的颜色特征 ;利用汽车上的反射光或阴影区域的亮度特征来进一步确认辨识结果 ,以克服复杂环境中背景颜色和各种光照视角、强度的变化对图像特征所产生的影响。此外 ,文中还提出一种新颖而快速的自适应“移位滤波器”算法 ,以消除颗粒状和细长带状噪声 。
This paper presents a real time image processing system, with a stereo color image instrument being chained to Matlab platform, for automatically identifying automobiles ahead and then calculating the distance between the instrument and the object A fuzzy rule based on a color model (hue saturation value) is proposed to extract the color characteristic of the taillights, motivated by the fact that all the automobiles have a pair of red color taillights and the HSV model conforms to the color consistency of human vision system In order to accommodate varying background colors and illumination conditions, the rear window or the shadow of a automobile is further identified by using the value information in the HSV model Moreover, a new algorithm is presented here for quick filtering out the noise consisted of the granulated or slim pixels Finally, experimental results are reported and analyzed.
出处
《国防科技大学学报》
EI
CAS
CSCD
北大核心
2002年第1期100-104,共5页
Journal of National University of Defense Technology
