Citation

黄, 黎, 冯, 梁, 伍. Comput. Digit. Eng. 2017; 2017(03): 523-526+564.

Vernacular Title

公共场所暴力行为的光流场构建及分析方法

Copyright

(Copyright © 2017, 中国船舶重工集团公司第七0九研究所)

DOI

unavailable

PMID

unavailable

Abstract

In view of the situation that in recognition of violence in public places, the previous method of detecting has the existence of miscarriage of justice to the static object or general physical contact. So in this paper,a public places violence detection method based on Lucas-Kanade optical flow method is proposed, and the effect of recognition in different parameters is analyzed. First, the feature points of two adjacent frames are detected and stored. Then,the optical flow field to reflect the motion state of objects will be constructed. Besides, in this paper, the effects of the three parameters on the optical flow are analyzed, which are the Pyramid decomposition layer, the optical flow computation window and the maximum optical flow direction of the Lucas-Kanade optical flow algorithm. The experimental results show that the analysis method for constructing optical flow field can effectively identify the occurrence of violence in public places.

Language: zh

Vernacular Abstract

针对以往在暴力事件识别中经常存在将相对静止的对象或普通身体接触误判为暴力行为的情况,论文提出了一基于Lucas-Kanade光流法的公共场所暴力检测方法。首先,检测并存储相邻两帧图像的特征点,其次,根据特征点对运动物体进行运动特征提取,最后,构建反映物体运动状况的光流场。同时,论文还分析了Lucas-Kanade光流算法的金字塔分解层数、光流运算窗口、最大光流方向数这三个参量对光流效果的影响。实验结果表明,使用构建光流场的分析方法能有效识别公共场所中发生的暴力事件。