文献类型：期刊文献

英文题名：The algorithm optimization based on the action recognition of silhouette and optical flow features

作者：Jia, Hailong[1]

第一作者：贾海龙

通讯作者：Jia, H.

机构：[1] Center of Modern Education Technology, Xinxiang University, Xinxiang 453000, China

第一机构：新乡学院

通讯机构：[1]Center of Modern Education Technology, Xinxiang University, Xinxiang 453000, China|[11071]新乡学院;

年份：2014

卷号：32

期号：2

起止页码：1231-1238

外文期刊名：Energy Education Science and Technology Part A: Energy Science and Research

收录：EI(收录号：20142317785747);Scopus(收录号：2-s2.0-84901475493)

语种：英文

外文关键词：Algorithms - Gesture recognition - Motion estimation

摘要：Human action recognition accuracy will be affected by a person's appearance, the environment, camera settings and other factors. Choices of the movement characteristics directly affect the recognition effect of human action recognition methods. Single feature always brings about the limited effect due to its different scope of application. Based on the study of the surface features and recognition of human action, this paper, after fully considering the advantages and disadvantages of different characteristics, proposes a mixed feature which combines the global silhouette features and local optical flow features, and applies it to the human action recognition. The experimental data show that the algorithm of this paper greatly improves the recognition accuracy more than the previous algorithms. ? Sila Science.