Image emotional semantic annotation based on fusion features
Xuliang Zhang, Sudi Lou · 2017
Due to “semantic gap”, the problem of image emotional semantic annotation has not been solved. In this parper, a method of emotion semantic annotation for cheongsam images based on Fusion Features has been proposed. Multi-features including the color and texture are used to describe the content of the image. Then least squares support vector machine for regression which is optimized by particle swarm optimization is used to build the mapping between the feature space and emotional space. The experiment indicates that this method achieves good effect.