Negatif olmayan matris ayrıştirma ile ikili video kiyimlama binary video hashing by non-negative matrix factorization

Abstract

This paper presents the video fingerprinting method proposed to TRECVID Content Based Copy Detection task by ITU MSPR Group. The proposed system consists of two modules: video fingerprint extraction and search-retrieval. Fingerprint extraction module performs a binary hashing scheme to generate the video fingerprints by using the mixing and encoding matrices obtained by non-negative matrix factorization (NMF) which is commonly used in dimension reduction. Search and retrieval module first applies a hash matching procedure to locate potential matching points. It is followed by temporal merging that eliminates false alarms while combining subsegments. Test results obtained on TRECVID video sequences illustrate that the proposed fingerprinting method is highly robust to insertion of pattern, blurring, change of gamma and noise addition attacks. We are currently working on to improve the robustness to geometric transformations such as shift, crop, flip andpicture-in-picture.