@inproceedings{b10e5814635b4ef4860592d71dc2332f,
title = "Research of small parts gesture estimation based on multilevel RVM regression",
abstract = "As to the real-time positioning demands for micro assembly process, this paper proposes a way which is based on Relevance Vector Machine Regression (RVMR). It solves the low efficiency problem which usually accompanies other common regression algorithms because the regression pattern is not sparse enough. This paper brings out grading RVMR, adopting the thought what is called 'From coarse to fine'. In this way, the number of training samples is greatly reduced while guaranteeing precision. So the off-line training efficiency is improved, meeting various parts in micro assembly process. In this algorithm, the algebra feature of the part image is extracted as the RVM's input, using Principal Component Analysis (PCA). Experiments on many regression algorithms and grading RVMR are both carried on. The results show that RVMR gets the shortest measuring time and the highest accuracy. The single axis estimation precision of part attitude is better than 0.5°.",
keywords = "Attitude estimation, Micro assembly, Regression, Relevance vector machine",
author = "Xiaojun Chen and Tao Hu and Dandan Wang and Huilan Wu",
year = "2013",
doi = "10.1109/ICEMI.2013.6743161",
language = "英语",
isbn = "9781479907571",
series = "Proceedings of 2013 IEEE 11th International Conference on Electronic Measurement and Instruments, ICEMI 2013",
pages = "877--881",
booktitle = "Proceedings of 2013 IEEE 11th International Conference on Electronic Measurement and Instruments, ICEMI 2013",
note = "2013 IEEE 11th International Conference on Electronic Measurement and Instruments, ICEMI 2013 ; Conference date: 16-08-2013 Through 18-08-2013",
}