摘要
在车载智能视觉采集系统的研究中,由于当前车载智能视觉采集过程对车道上行人距离的判断没有考虑夜视环境的特殊性,也没有考虑图像视觉处理过程中由于夜间带来的计算误差对行人距离计算的影响,判断的行人距离在夜间存在较大误差。为了避免上述弊端,提出了一种考虑夜视误差补偿的车道行人距离计算方法。通过分析夜视图像中的距离计算特征,运用一种插值误差补偿的方法,对由于夜视原因造成的像素过于宽泛误差进行补偿,能够将夜视环境下的视觉误差进行补偿,将不同坐标系统中的车道与行人点进行转换,获取误差补偿后的车道坐标系统中对应点的空间位置和行人坐标系统中对应点的空间位置,实现精确计算。实验结果表明,利用改进后的算法进行车道行人距离计算,能够有效提高识别的准确性。
In the study of intelligent vision - board acquisition system, the special nature of night vision environ- ment is not considered in the distance of the pedestrian lane determination in current vehicle intelligent visual acquisition process. Moreover, the calculation error caused by the night is also not considered in the distance calculation of pedestrians in the visual image processing. Thus, there is a big error for pedestrians distance judgment at night. To avoid these drawbacks, this paper presented a pedestrian lane distance calculation method which considers the night vision error compensation. Firstly, by analyzing the characteristics of the image, the distance night vision was calcu- lated. A method of interpolation was used to make error compensation caused by the night vision, which can be visu- ally compensated for night vision environment. Then the pedestrian and lane point in different coordinate systems was converted to obtain the spatial position of the corresponding point in lane coordinate system and pedestrian coordinate system after error compensation and achieve accurate calculations. Experimental results show that the improved algo- rithm applied in the distance calculation between pedestrian and lane can improve the recognition accuracy.
出处
《计算机仿真》
CSCD
北大核心
2014年第8期159-162,共4页
Computer Simulation
基金
江西省科技支撑计划项目(2011ZBBE50029)
关键词
车道行人
距离计算
误差补偿
Lane pedestrians
Distance calculation
Error compensation
