CAJ | 학술논문

在基于视频的运动目标检测算法中，背景差分法是一种常用的算法，可是由于时间推移，外部光照和阴影发生变化，会导致背景不断发生变化，因此背景的更新是很重要的一部分，直接影响到目标检测的提取效果。该文提出了基于卡尔曼滤波的背景更新算法，实时实现背景更新，通过滤波器分出运动目标扰动，得到背景模型序列。算法简单，能较好的处理光照变化，有效的实现背景更新。
재기우시빈적운동목표검측산법중，배경차분법시일충상용적산법，가시유우시간추이，외부광조화음영발생변화，회도치배경불단발생변화，인차배경적경신시흔중요적일부분，직접영향도목표검측적제취효과。해문제출료기우잡이만려파적배경경신산법，실시실현배경경신，통과려파기분출운동목표우동，득도배경모형서렬。산법간단，능교호적처리광조변화，유효적실현배경경신。
Based on the video moving target detection algorithms, the background difference method is a commonly used algo-rithm, but with the passage of time, the light and the external conditions will make the true background is constantly changing, so the background updating is the most important part, directly affect the target detection and extraction effect. This paper presents Calman filtering background estimation algorithm based on low pass filter is used, and the moving object is separated from back-ground disturbance, model sequence. The algorithm is simple, can be used to deal with illumination changes, the effective imple-mentation of background updating.