The following information was submitted:
Transactions: WSEAS TRANSACTIONS ON SIGNAL PROCESSING
Transactions ID Number: 32-288
Full Name: zuojin li
Position: Professor
Age: ON
Sex: Female
Address: No.174, Shazheng Street, Shapingba District, Chongqing
Country: CHINA
Tel:
Tel prefix:
Fax:
E-mail address: lizuojin110@tom.com
Other E-mails: cqulzj@gmail.com
Title of the Paper: Independent Component Analysis and Recognition of Natural Images
Authors as they appear in the Paper: Zuojin Li, Weiren Shi, Zhi Zhong
Email addresses of all the authors: lizuojin110@tom.com,wrs@edu.cqu,zzhong@mae.cuhk.edu
Number of paper pages: 10
Abstract: Natural images recognition is an important area of machine vision. This paper presents a novel approach for natural images recognition, based on the non-Gaussian distribution property of natural images. In this new method for recognition, first supervised classification is conducted to the natural images based on their label value, then independent components linear transforms are conducted to each category of samples, high-dimensional data are transformed to irrelevant independent components, and finally the probability distance between independent component subspaces is used for unsupervised classification. This classification tree also shows some features of signal processing of biological optic nerve. Experiment on ORL Face Database identity recognition shows that this method is featuring high recognition rate and low time consumption; meanwhile, another experiment is conducted on direction determination of intelligent robots autonomous navigation, also produci!
ng a good result.
Keywords: Natural image, Independent component subspace, Hierarchical discriminant regression, Recognition of face£¬Robot navigation
EXTENSION of the file: .pdf
Special (Invited) Session:
Organizer of the Session:
How Did you learn about congress:
IP ADDRESS: 123.147.107.143
