CAJ | 학술논문

为提高全天候复杂光线条件下高清视频监控系统中遗留物检测的实时性、准确性和鲁棒性，提出全天候复杂光线条件下的遗留物实时检测方法。采用快速的混合高斯模型算法对视频图像进行背景建模，分别建立一个长周期背景模型和一个短周期背景模型，通过当前视频帧分别与两个背景模型差分运算，得到长周期前景和短周期前景；对长周期前景和短周期前景进行分析，检测标记出遗留物，进行报警。实验结果表明，该方法能有效检测出复杂光线条件下的遗留物，具有较高的实时性和准确性。
위제고전천후복잡광선조건하고청시빈감공계통중유류물검측적실시성、준학성화로봉성，제출전천후복잡광선조건하적유류물실시검측방법。채용쾌속적혼합고사모형산법대시빈도상진행배경건모，분별건립일개장주기배경모형화일개단주기배경모형，통과당전시빈정분별여량개배경모형차분운산，득도장주기전경화단주기전경；대장주기전경화단주기전경진행분석，검측표기출유류물，진행보경。실험결과표명，해방법능유효검측출복잡광선조건하적유류물，구유교고적실시성화준학성。
To improve the timeliness ,accuracy and robustness of abandoned object detection under the condition of complex ligh‐ting around the clock in the HD video surveillance system ,an abandoned object detection method under the condition of complex lighting around the clock was presented .Firstly ,a fast mixed Gaussian modeling method was adopted ,and two background models based on two different update rates were established respectively .The current video frame was subtracted from the two background models ,and two foreground models were obtained .Secondly ,the abandoned object was detected and marked after analyzing two foreground models .Experimental results show that the proposed method can effectively detect the abandoned ob‐ject under the condition of complex lighting around the clock ,and it has good performance in timeliness and accuracy .