Indoor and outdoor image classification with bilateral filter

With the exponential increase in the number of photos, the operations such as organization, browsing and searching etc. are getting more and more difficult. Image category based on the content is significantly helpful to these tasks. In various kinds of image classification, the classification of indoor and outdoor images is fundamental. In this paper, we propose a method for indoor/outdoor image classification that can achieve better accuracy without slowing down the computation speed. It is based the view that the main differences between indoor and outdoor images are the light sources and material of objects. The operation of data set training and image classification are done with the information of light sources and objects´ material which can be approximately obtained with a revised bilateral filter. Experimental results show that the proposed scheme is capable of classifying indoor/outdoor images with better accuracy at a reasonable speed.