Hidden Markov model based video indexing with discrete cosine transform as a likelihood function
Maheshkumar H. Kolekar, Somnath Sengupta · 2005
In this paper, we have proposed discrete cosine transform (DCT) as a likelihood function for video shot detection and classification using hidden Markov model and experimentally observed that only DC coefficient of DCT performs better with heavily reducing computational burden. Results of DC-DCT likelihood function are presented and found encouraging.