Screening of Tobacco's Effective Image Features Based on a Semi -supervised Clustering
Ge Jin, Lin Qi, Hang Li
Available Online December 2016.
- https://doi.org/10.2991/mcei-16.2016.59How to use a DOI?
- Clustering; Discriminant function; Feature selection; Support vector machine; Tobacco classification
- In order to reduce the number of extracting tobacco's image features and the computational complexity of hierarchical model, and to increase the speed and accuracy of tobacco classification, this paper presents a feature selection method that based on the semi - supervised clustering. First, define the discriminate function R which can distinguish good features from bad features, and delete bad features according to the R-value. Then set up 42 levels' SVM hierarchical model using all features and screened features. Experimental results show that the feature selection method constructed in this paper can select effective features, which can raise the tobacco's classification speed under the premise of correct discrimination.
- Open Access
- This is an open access article distributed under the CC BY-NC license.
Cite this article
TY - CONF AU - Ge Jin AU - Lin Qi AU - Hang Li PY - 2016/12 DA - 2016/12 TI - Screening of Tobacco's Effective Image Features Based on a Semi -supervised Clustering BT - 2016 6th International Conference on Mechatronics, Computer and Education Informationization (MCEI 2016) PB - Atlantis Press SP - 281 EP - 284 SN - 1951-6851 UR - https://doi.org/10.2991/mcei-16.2016.59 DO - https://doi.org/10.2991/mcei-16.2016.59 ID - Jin2016/12 ER -