Scene classification is one of the most appealing and challenging problems in computer vision. Recently, bag-of-visual-words (BOVW) method using spatial pyramid matching scheme has shown remarkable performance for scene classification. But this method deriving from local keypoints does not contain texture features which are rich in scene images. To further improves the classification accuracy, this paper presents a scene classification method combining rotation invariant local binary patterns (RILBP) texture features and bag-of-visual-words in spatial pyramid matching framework. First, scene image is subdivided at three different levels of resolution for constructing a spatial pyramid. Then based on scale invariant feature transform (SIFT) descriptor and K-means clustering, pyramid histogram of visual words is extracted. Pyramid histogram of RILBP texture features is extracted using the mean of a 3*3 neighborhood as threshold. Last we construct a composite kernel of spatial pyramid matching. We regard the keypoint features and texture features as two independent feature channels, and combine them to realize scene classification using one-against-all SVMs with the composite kernel. Experiments on the three different scene datasets demonstrate the effectiveness of the proposed method.