利用人眼感知视觉模型的车型动态定位

Dynamic vehicle localization based on visual perception model

通过对实时汽车图像的大量研究,提出基于人眼感知视觉模型的自适应汽车车体定位算法。在HSI彩色图像空间,根据人眼分辨率的视觉感知模型分别在色度、亮度和饱和度空间进行自适应的边缘检测;将3个空间得到的图像边缘合并以得到整个车体在实际影像序列中的边缘;最后利用投影知识原理,分别利用边缘的垂直投影图和水平投影图来定位车体的左右边界和上下边界。实验结果表明,该算法具有较快的速度和较好的准确性,能满足实时的车型识别系统的要求。

Based on human visual perception model technique, a novel and intuitive method for location vehicle is presented. The perception color space adopted is HSI color space. Three color components of a color image and more potential edge patterns are integrated for solving the feature extraction problem. A fast and automatic threshold technique based on human visual perception model is also developed. Vertical edge projection and horizontal edge projection are used to locate the leftright boundary and topbottom boundary of vehicle, respectively. Experimental results show that proposed location vehicle method is efficient and reliable, and can satisfy the demands of vehicle recognition system.