A scene image recognition method based on extension features

To improve the performance of scene image recognition, new recognition methods based on extension features are presented. Using expansion and contraction transform, a new extension color histogram feature E-hist and related recognition method are designed. Then, another new extension feature E-SURF is designed based on divergence analysis. The corresponding recognition method is also proposed with displacement transformation of extension theory. Large quantities of experiments show that the two extension features work well and the presented recognition method can improve scene image recognition accuracy effectively.