Abstract— In this paper, a new technique for keyframe extraction and shot boundary detection is proposed. The proposed technique uses Eigen values for measuring the dissimilarity between the consecutive frames. The behavior of the Eigen values of covariance matrix over certain frames is analyzed to detect the shot boundary and subsequently extract the keyframe. In this paper a novel method for dynamically updating the covariance matrix is also proposed. This greatly reduces the computational time compared to conventional method of calculating the covariance matrix. The proposed technique is tested on a large variety of videos and also works on live videos. The recall and precision of simulation results reveal that the proposed technique is highly efficient and accurate in detecting the abrupt cuts.
Index Terms— Computational efficiency, covariance matrix, dynamic update, Eigen values, keyframe extraction, Shot boundary detection.
Vijeetkumar Benni, P. Punitha, and Vasudeva Rao are with Imaging Tech Lab, HCL Technologies, Bangalore, India (e-mail: vijeetkumar.Benni@hcl.com, drpunitha.s@hcl.com, vasudeva.r@hcl.com).
R. Dinesh was with Imaging Tech Lab, HCL Technologies, Bangalore. He is now with Amazon, Bangalore, India (e-mail: dr.dineshr@gmail.com).
[PDF]
Cite: Vijeetkumar Benni, R. Dinesh, Punitha P., and Vasudeva Rao, " Keyframe Extraction and Shot Boundary Detection Using Eigen Values," International Journal of Information and Electronics Engineering vol. 5, no. 1, pp. 40-45, 2015.