Privacy preserving in CBIR using color and texture features
Mukul Majhi, Sushila Maheshkar · 2016
The advent of digital technology and its range of applications in various field witness the importance of retrieving huge and diverse multimedia content from the data repositories. The contents are always under potential threat intended to extract the private information. This paper presents a privacy preserving technique to retrieve images from the corresponding database by using the encrypted feature vector. Texture and quantized HSV color space histogram features are exploited to formulate the feature vector which is encrypted by performing XORed operation with its sliced biplanes by a random binary bit pattern to preserve the hamming distance. Finally, random permutation provide the encrypted feature vector. Experimental results illustrate that the method preserve private information of the content and retrieve relevant images effectively and efficiently.