Abstract: An optimization method is proposed for moving object detection in image sequence. As to the existence of “pseudo object” noise when the different method for indoor object detection is applied and “vestigial-image” noise problem while background estimation method for outdoor object detection is used on for short sequences, this paper finds and employs the relativity between twice difference to detect and eliminate the “pseudo object”, and then combines difference and background estimation method to eliminate the “vestigial-image” noise that cannot be avoided when only the latter is used. Experiment results show that, the proposed method can not only eliminate the “pseudo object” and “vestigial-image” noise, but also help to improve detection integrity remarkably.

Key words: image sequence, object detection, image differences, background estimation

摘要： 提出了一种用于图像序列中检测运动目标的优化算法。针对用于室内目标检测的差分法存在着“虚影”噪声，以及用于室外目标检测的背景估计法在对短序列进行检测时，其结果中存在“残像”噪声的问题，揭示并利用两次差分之间的相关性实现了对“虚影”的检测并将其消除，将其引入背景估计法，以消除后者存在的“残像”噪声。实验表明，该方法在目标检测中不仅消除了“虚影”和“残像”噪声，而且检测结果的完整性显著提高。

关键词: 图像序列, 目标检测, 差分法, 背景估计法