Automatic Season Classification of Outdoor Photos

Automatic season classification of an image is potentially useful for content-based image retrieval, computer object recognition and digital photo management applications. In this paper, we propose a method for automatic season classification of outdoor photos combining color information and skin information. We extract color information by computing the normalized color histogram of the photo. Skin information is only available for the photos containing people. It is extracted based on the results of face detection and skin detection. After performing face and skin detection, we select skin blobs that are probably face or limbs based on local feature analysis. Then features are extracted from the selected skin blobs to represent the information of skin exposure. Color information and skin information are combined using the Bayesian method. Experimental results have shown the effectiveness of the proposed method.