English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Human action recognition is a highly active research topic in the field of video surveillance, human computer interaction and other fields. Due to the huge amount of computation, many existence methods fail in real-time applications. In this paper, we proposed a human action recognition method based on human skeleton Hu invariant moments combined with human geometrical characteristics. Firstly, foreground is extracted by the inter-frame difference and background subtraction. Secondly, human skeleton Hu invariant moments, the minimum bounding rectangle aspect ratio, rectangularity and circularity are calculated. Finally, human actions are recognized by the K-nearest neighbor. Experimental results in Weizmann action datasets show that this method can accurately identify human actions and has good real-time performance. The proposed method can be applied to real-time intelligent video surveillance with high accuracy. Introduction Human action recognition has been attracted much attention over the past few decades due to its applicability to many areas, including video surveillance, human computer interaction, etc. Many scholars have concentrated on the accuracy of human action recognition and contributed multiple proven methods. Nguyen et al. [1] build a human body model with shape template matching to detect the moving human. When complex background occurred, it may lead to identification errors. Min et al. [2] use an action region-aware dictionary to identify human actions. Lin et al. [3] propose an effective graph model based on Action Trait Code classification to recognize human actions. The previous methods are high recognition rates which cost complex algorithms and large amount of computations. Considering the running time and validity of the algorithm, we research a human action recognition method based on human skeleton Hu invariant moments combined with human geometrical characteristics. Inter-frame difference and background subtraction are combined to detect moving objects. Feature extraction based on human skeleton Hu invariant moments and human geometrical characteristics. K-nearest neighbor classification is used to identify human actions. This method is real-time and fast in execution time. It can be used in real-time alarm. Implementation of Human Action Recognition In this paper, we use human action recognition method which based on human skeleton Hu invariant moments combined with human geometrical characteristics. This method includes moving object detection, human action feature extraction and human action recognition. The diagram of human action recognition algorithm is shown in Fig.1 Firstly, Gaussian filter and the expansion of corrosion closing operation are used to remove the noise of video sequence. Secondly, a combined method which includes inter-frame difference, background subtraction and background updating model [4] is 1628 5th International Conference on Frontiers of Manufacturing Science and Measuring Technology (FMSMT 2017) Advances in Engineering Research (AER), volume 130

Human Action Recognition based on Human skeleton Hu invariant moments combined with human geometrical characteristics