A scene recognition algorithm based on covariance descriptor

Covariance descriptor which is a good method for object detection represents a region of interest by using the covariance of d-features. Different scene images are regarded as textures with different labels in this paper, and then a novel scene recognition algorithm based on covariance descriptor is proposed. Firstly, the covariance feature of scene image is retrieved. Secondly, support vector machine is employed for training the combination classifiers which recognize the scene images. The experimental results demonstrate the feasibility and validity of the proposed algorithm.