Scene classification based on single-layer SAE and SVM

作者：

Highlights：

• The SAE and SVM based scene classification method achieves excellent performance.

• A single-layer SAE is proposed to extract sufficient and appropriate features.

• A mean-pooling operation is proposed to reduce the dimension of feature vectors.

• A PSO-based algorithm is employed to optimize the parameters for the SVM classifier.

摘要

•The SAE and SVM based scene classification method achieves excellent performance.•A single-layer SAE is proposed to extract sufficient and appropriate features.•A mean-pooling operation is proposed to reduce the dimension of feature vectors.•A PSO-based algorithm is employed to optimize the parameters for the SVM classifier.

论文关键词：Scene classification,Feature learning,Single-layer SAE,SVM,PSO

论文评审过程：Available online 19 December 2014.

论文官网地址：https://doi.org/10.1016/j.eswa.2014.11.069