Encrypted Image Retrieval System: A machine learning approach

The purpose of this paper is to describe our research and solution to the problem of designing an Encrypted Image Retrieval System. Due to rapid increase in image database volume, as well as its vast deployment in various applications, the need to secure images in the database became necessary and the need for development of efficient algorithms for proposed retrieval systems are important. The proposed solution in the paper outlines the encryption algorithm, the feature database construction algorithm and the image retrieval mechanism. The feature database is constructed by obtaining certain features of image like colour moments, HSV histograms which are key features for identification of colour distribution if an image. The proposed solution also includes calculating certain space frequency localization coefficients of the image like wavelet moments and Gabor coefficients block wise. The central tendency measure like mean, variance and standard deviation of the above mentioned space frequency localization coefficients are stored as the part of feature vector which helps for retrieving images using supervised learning algorithms. The extracted features are used for similarity check between images using K Nearest Neighbour algorithm and Support Vector Machine algorithm.