论文部分内容阅读
通过对实际路口昼夜路面状况的分析,提出在“视频电子警察系统”中进行昼夜区分以提高车辆检测精度的必要性,分析了常见的昼夜区分方法的优缺点,在此基础上,研究提出一种利用路面灰度特征,对视频图像进行统计学习进行昼夜区分的算法,通过应用大量实测图像数据,得出了相应的统计数据曲线,验证了算法的可靠性、高效性以及通用性.
Based on the analysis of the condition of day and night pavement at the actual junction, the necessity of diurnal and night discrimination in video electronic police system is put forward to improve the detection accuracy of vehicles. The advantages and disadvantages of common day-night classification methods are analyzed. Based on this, According to the grayscale feature of pavement, the algorithm of day-night distinction of video image is studied statistically. Through the application of a large amount of measured image data, the corresponding statistical data curve is obtained, which proves the reliability, high efficiency and universality of the algorithm.