AN ENHANCED THREE-STAGE FOREGROUND ANALYSIS AND TRACKING ALGORITHM FOR BLUE SCREEN COMPOSITION
One of the most prevalent approaches to identifying video forgery is the use of blue-screen composition. However, from literature, very limited algorithms exist for detecting video forgery. An enhanced three-stage Foreground Analysis and Tracking (E3FAT) algorithm has been developed to detect blue-screen composition. The E3FAT framework operates in three phases: In the first stage, foreground blocks are taken from the target video using a Gaussian Mixture model (GMM). During the second stage, the homogeneity function is utilized to process the extracted images taken from the target video. In the final phase, forged blocks are rapidly tracked using the Discriminative Correlation Filter with Channel and Spatial Reliability (CSR-DCF) algorithm. Empirical evaluation demonstrates that E3FAT achieves a reliable detection of video manipulation, of 98.03% true positive rate and an average processing time of 95.57 seconds.
Keywords: Blue Screen Composition, Homogeneity, Foreground Analysis, Three-Stage. Tracking algorithm
