CAJ | 학술논문

在计算机视觉应用领域，视频图像的目标检测、跟踪以及压缩，经常需要检测运动目标，运动目标检测的难点在于复杂图像背景条件下对运动目标的检测与分割。因此一种能够自动检测与准确分割运动目标的算法对于应用系统的性能具有重要意义。提出了一种基于后验概率的运动目标检测算法，在概率模型中采用了时域和空域边缘特征，通过在视频图像上进行的实验证明，该算法有较好的分割运动目标效果。
재계산궤시각응용영역，시빈도상적목표검측、근종이급압축，경상수요검측운동목표，운동목표검측적난점재우복잡도상배경조건하대운동목표적검측여분할。인차일충능구자동검측여준학분할운동목표적산법대우응용계통적성능구유중요의의。제출료일충기우후험개솔적운동목표검측산법，재개솔모형중채용료시역화공역변연특정，통과재시빈도상상진행적실험증명，해산법유교호적분할운동목표효과。
In computer vision field, the target detecting, tracking and compressing of video image requires detecting the moving target. The difficulty is the detecting and segmenting of moving target detection in complex background. So it is very important to find out an automatic and accurate segmenting algorithm for the performance of an application. A improved moving target detection algorithm based on maximum posterior is proposed. The edge characters in time and space region are adopted in the proposed algorithm, through the experiments on video images; the algorithm is proved having better moving target segmenting effects.