Context Based Image Retrieval using Image Features

Context based image retrieval is useful to effectively retrieve images based on its context from the massive collection available on the web. Content based image retrieval system is effective only to recognize and retrieve specific objects. It fails when the image has multiple objects and the objects are cluttered. In this work an algorithm has been proposed which facilitates the search based on the context,”arrangement of an object with respect to other objects” in a given image. Image containing the objects with its context is chosen as input, images which have the chosen object with the similar context have been retrieved from the dataset. Supervised segmentation and labelling of an image has been done. Features such as Histogram Orientation of Gradients (HOG), Scale Invariant Feature Transform (SIFT), 1-D signature, Gray Level Co-occurrence Matrix (GLCM) texture properties, Statistical properties (mean, standard deviation), Speeded Up Robust Features (SURF) and Colour Histogram are extracted from the image. Given a training set of images, context is learnt using the minimum distance classifier based on the extracted features. Study of this algorithm on the Kitchen Scene in the Scene UNderstanding (SUN 09) data set has been done. Image retrieval is done based on all the set of features extracted from the images. Context based image retrieval system has been widely used in object recognition, person identification within the context of personal photo collection and image annotation.