CAJ | 학술논문

文章提出一种基于背景减除法与三帧差分法相融合运动目标检测的方法，该算法运用改进的时间平均算法初始化背景，在有目标情况下也能构建出可靠的背景；用背景减除法得到目标运动区域，并根据像素的连通区域除噪声；通过背景减除法与三帧间差分的检测结果进行“与”运算，得到的结果作为反馈对背景模型进行更新；利用色度信息消除阴影，得到较精确的运动目标。实验结果表明，该算法能够快速、完整、准确地检测出运动目标，满足实时检测的需要。
문장제출일충기우배경감제법여삼정차분법상융합운동목표검측적방법，해산법운용개진적시간평균산법초시화배경，재유목표정황하야능구건출가고적배경；용배경감제법득도목표운동구역，병근거상소적련통구역제조성；통과배경감제법여삼정간차분적검측결과진행“여”운산，득도적결과작위반궤대배경모형진행경신；이용색도신식소제음영，득도교정학적운동목표。실험결과표명，해산법능구쾌속、완정、준학지검측출운동목표，만족실시검측적수요。
A detection algorithm of moving objection based on background subtraction and three-frame differen-cing is presented in this paper . T he algorithm uses the improved time-averaged method to initialize back-ground ,and can obtain reliable background in the condition of having moving object .The moving object re-gion is gained by using the background subtraction ,and the noise is suppressed according to the relationship between the pixel and its adjacent pixels .The“and”operation is done according to the detection results of the background subtraction and the three-frame differencing ,and the results obtained are used as feedback to up-date background .Finally ,the chromatic difference is employed to eliminate shadows and make the moving object region more accurate .T he simulation results show that the presented algorithm can detect the moving object quickly ,completely and accurately ,and is suitable for real-time detection .