| Summary: | The problem of indoor-outdoor image classification using supervised learning is addressed in this paper. Conventional indoor-outdoor image classification methods, partition an image into predefined sub-blocks for feature extraction. However in this paper, we use a simple color segmentation stage to acquire meaningful regions from the image for feature extraction. The features that are used to describe an image are color correlated temperature, Haralick features, segment area and segment position. For the classification phase, an MLP was trained and tested using a dataset of 800 images. A classification accuracy of 94% compared with the result of other state of the art indoor-outdoor image classification methods showed the efficiency of the proposed method.
|
|---|