Publication Type : Conference Paper
Publisher : Springer Singapore
Source : Communications in Computer and Information Science
Url : https://doi.org/10.1007/978-981-13-1936-5_46
Campus : Amritapuri
School : School of Computing
Department : Computer Science and Engineering
Year : 2018
Abstract : The images are having significant role in our daily life. Images consume more storage space when compared to text documents. For preserving privacy of images, before deploying it to cloud storage images are encrypted. A scheme supporting CBIR (content-based image retrieval) from encrypted images is proposed in this paper. The features are identified from the outsourced images and by applying locality-sensitive hashing pre-filter tables are generated for increasing the efficiency of searching. The features of the outsourced images are represented by using interest points and are encrypted by using a stream cipher. A machine learning algorithm using FAST method identifies the interest point on image contour, which helps in retrieving most similar images from cloud. Besides these, for avoiding illegal distribution of retrieved images by query, the cloud server embedds a unique watermark to the encrypted images by using a watermark based protocol. The average search time and precision of the proposed system can be inferred from performance evaluations.
Cite this Research Publication : N. Sharmi, P. Mohamed Shameem, R. Parvathy, Content-Based Image Retrieval Using FAST Machine Learning Approach in Cloud Computing, Communications in Computer and Information Science, Springer Singapore, 2018, https://doi.org/10.1007/978-981-13-1936-5_46