Title :
Human Action Classification Based on the 3D Zernike Moment
Author :
Hao Yan ; Zhu Zhenwen ; Lu JianFeng
Author_Institution :
Nanjing Univ. of Sci. & Technol., Nanjing, China
Abstract :
The Zernike moments are a set of orthogonal moments, which can describe more details of the target than the Hu moments. The 3D Zernike descriptors are natural extensions of spherical harmonics based descriptors. In this paper, we adopt the 3D Zernike moment to calculate the global feature of the human action, then, classify the image sequences with the Bayes based AdaBoost classifier. Our experiment results show that the 3D Zernike moment is better than geometric moment in human action classification.
Keywords :
Bayes methods; image classification; image motion analysis; image sequences; learning (artificial intelligence); 3D zernike descriptors; 3D zernike moment; Bayes based AdaBoost classifier; Hu moments; geometric moment; human action classification; image sequences; orthogonal moments; spherical harmonics based descriptors; Databases; Feature extraction; Poisson equations; Shape; Support vector machine classification; Three-dimensional displays; Training; 3D Zernike Moments; Action Classification; AdaBoost Classifier; Human Action;
Conference_Titel :
Measuring Technology and Mechatronics Automation (ICMTMA), 2014 Sixth International Conference on
Conference_Location :
Zhangjiajie
Print_ISBN :
978-1-4799-3434-8
DOI :
10.1109/ICMTMA.2014.76
