PERCEPTUAL VIDEO HASHING BASED ON WALSH HADAMARD TRANSFORM WITH APPLICATION TO INDEXING AND RETRIEVAL OF NEAR IDENTICAL VIDEOS

Asha kiran M U, Sandeep R

Abstract: Advancement in technology enables us to access videos at all times and even modify them with available media editing tools. This leads to fraudulency of the video and even the author copyrights are intruded. Hence, there is a need to identify the forgery of the video and exploitation of author copyrights. Many techniques have been described in the literature to overcome these problems; they are Cryptographic hashing, Watermarking and Perceptual hashing. Our main concern here is towards perceptual video hashing. Perceptual video hashing captures the perceptual gist of the video into a concise string called the perceptual hash. To counteract malicious modification and exploitation of copyrights of the video, obtained perceptual hash is employed. The hash is also used for video indexing and retrieval applications. The proposed method uses Walsh Hadamard Transform for the engendering of perceptual hash. Hash generated is used for authentication and video retrieval and indexing method. The algorithms performance is weighted on the basis of ROC curves and recall versus precision curves.

Keywords: Perceptual hash, Authenticity, Copyright and Walsh Hadamard Transform.

DOI: https://doi.org/10.15623/ijret.2017.0607011