Feature selection through adaptive sparse learning for scene recognition

Yunyun Sun, Peng Li, Hang Sun, He Xu, Ruchuan Wang

Abstract: Highlights•We propose an innovative adaptive sparse learning feature selection approach, aiming to increase inter-class discrimination and improve scene recognition accuracy.•Develop a sparse learning method to construct a salient feature dictionary that explicitly describes the importance of features.•Design an alternate updating adaptive weight variable optimization method automatically adjusts feature weights to select common features.•Extensive experiments demonstrate the superiority and reliability of the proposed method in scene recognition tasks.