A novel video scene change detection using successive estimation of statistical measure and HiBiSLI method

摘要

Video data watermarking becomes very essential due to digital broadcasting. Scene change detection is the initial step, required for the digital video data copy protection. This paper presents novel scene change detection algorithm based on successive estimation of statistical measure [SESAME] and HiBiSLI algorithm. The designed system detects the scene change using, covariance difference between the measures of each frame is considered. Successive difference between the statistical measures of frames are used and the difference is compared with a pre-difference value. A new algorithm, a combination of histograms, binary search and linear interpolation [HiBiSLI] filtering algorithm is introduced to filter out similar frames at the initial stage. The filtering method uses calculated threshold value. Scene detection algorithm uses ffmpeg as reference to calculate different parameter values. The performance of the designed system is evaluated using precision, recall, F-measure values and computation time. The experimental result proves that the designed system is well good in terms of its computational time, low complexity and better parameter values.