Aircraft type classification based on an optimized Bag of Words Model

摘要

Bag of Words model is used widely in image classification fields. In this paper, first, the traditional Bag of Words is introduced and used for aircraft type classification. Second, an optimized Bag of Words Model, which takes the place of the traditional Bag of words based on ordinary SIFT sampling and K-means clustering, is proposed based on space partition SIFT sampling and FCM clustering, and then is applied for recognizing aircraft types. Experimental results show that the optimized Bag of Words can keep higher recognition rate for aircraft type classification than the traditional Bag of Words and Affine Moments whatever origin aircraft images and add-noise aircraft images.