Moving object detection from video sequence using local motion direction histogram features

摘要

In this paper, we considered moving object detection technique from a video sequence that aimed at the content description of the video. For the video retrieval by the keyword, we must beforehand describe names and operations of the object in the video by linguistically information. In the reason, it is necessary to separate the object in the video from the background. The moving object detection technique is one of the most important technologies of the sake. The moving object detection has been studied as an important problem in fields such as the video monitoring. However, the problem of the camera work that appeared in case of the general video was not considered. For this problem, we found that removal of the background and detection of the camera work was possible by using local motion direction histogram features. We confirmed the effectiveness of our method through experimental results.