PROJECT TITLE :

DPcode: Privacy-Preserving Frequent Visual Patterns Publication on Cloud

ABSTRACT:

These days, cloud has become a promising multimedia data processing and sharing platform. Several institutes and corporations plan to outsource and share their giant-scale video and image datasets on cloud for scientific analysis and public interest. Among numerous video applications, the discovery of frequent visual patterns over graphical knowledge is an exploratory and vital technique. But, the privacy issues over the leakage of sensitive information contained in the videos/pictures impedes the further implementation. Although the frequent visual patterns mining (FVPM) algorithm aggregates outline over individual frames and appears not to pose privacy threat, the personal data contained in individual frames still might be leaked from the statistical result. During this paper, we study the problem of privacy-preserving publishing of graphical knowledge FVPM on cloud. We tend to propose the primary differentially non-public frequent visual patterns mining algorithm for graphical data, named DPcode. We tend to propose a completely unique mechanism that integrates the privacy-preserving visual word conversion with the differentially private mechanism underneath the noise allocation strategy of the sparse vector technique. The optimized algorithms properly allocate the privacy budgets among totally different phases in FPM algorithm over pictures and cut back the corresponding information distortion. Extensive experiments are conducted based mostly on datasets commonly employed in visual mining algorithms. The results show that our approach achieves high utility while satisfying a sensible privacy demand.