基于多颜色空间信息融合的视频车辆检测方法

利用图像颜色空间各分量的相对独立性和互补性,通过理论分析和实验,提出了有选择地将RGB、YCbCr和HSI3个颜色空间9个分量的检测结果进行融合的方法,提高了车辆视频检测的准确性。该法在S和I分量较大时依据H分量的值选择不同3个分量的检测结果进行有差别融合,而S和I分量较小时再选择另外3个分量的检测结果融合。实验证明,同只利用亮度分量的检测结果相比,检测到的车辆更加充实,空洞大幅度减少,很大程度提高了依赖于该结果的交通信息数据的精确性。

Video Vehicle Detection Method Based on Multiple Color Space Information Fusion

According to the components’ relative independence and complementariy in the image color space,a detection method has been presented after theoretical analysis and experiment which selectively fuses the detection results of nine components in RGB,YcbCr and HSI color spaces,thus to improve vehicle video detection accuracy.This method selects the detection results of three components to make different fusions according to the value of Component H when the value of S and I are both high,and another three components detection results will be done when S and I are both small.Experiments show that in contrast to the detection results using only brightness,the detected vehicles are more complete with greatly reduced vacant spots,which to a large extent increases the accuracy of the results concerned traffic information data.