Non-negativity and locality constrained Laplacian sparse coding for image classification

作者：

Highlights：

• Non-negativity for dictionary and codes is proposed to avoid features offsetting each other.

• Locality for sparse codes is added, which preserves local information for features.

• Laplacian regularization is used to preserve the consistency of codes with similar features.

• Our proposed method has better performance than previous algorithms.

摘要

•Non-negativity for dictionary and codes is proposed to avoid features offsetting each other.•Locality for sparse codes is added, which preserves local information for features.•Laplacian regularization is used to preserve the consistency of codes with similar features.•Our proposed method has better performance than previous algorithms.

论文关键词：Non-negativity,Locality,LSC,K nearest neighbourhoods,SPD,MP

论文评审过程：Received 11 August 2016, Revised 22 October 2016, Accepted 8 December 2016, Available online 12 December 2016, Version of Record 21 December 2016.

论文官网地址：https://doi.org/10.1016/j.eswa.2016.12.012