A proposed discriminative dictionary pair learning algorithm for image classification
Abstract
Dictionary learning for sparse coding has been widely applied in the field of computer vision and has achieved promising performance. In this paper, a new method called discriminative dictionary pair learning (DDPL) for image classification was proposed which jointly learned a synthesis dictionary and an analysis dictionary to promote the image classification performance. The DDPL method ensures that the learned dictionary has the powerful discriminative ability and the signals are more separable after coding. Compared with previous dictionary learning methods, DDPL employs projective coding, which largely reduces the computational burden in training and testing. Experimental results on various image classification benchmarks are presented to demonstrate the effectiveness of the proposed method.
Moraref