Frame duplication/mirroring detection method with binary features

A new video forgery detection approach based on forgery line

“…The integer-valued points on the line indicate the corresponding index values of forged frames. Comparison and effectiveness tests, given in the experimental result section, show that the proposed method gives better execution time than [1][2][3] and better detection capability when compared to similar works [1][2][3][4].…”

“…The average execution time results for the methods indicate that the proposed method can detect forgery operations on the test videos with less time. While the average per frame execution time for the methods in [1][2][3][4] is approximately 294.6, 140.6, 0.4, and 0.063, the proposed method can detect forgery operation on a frame within 0.07 s. Enhancing per frame execution time in video processing is important, because test videos can accommodate many frames, especially when surveillance videos are considered. Another experiment gives comparison results for the methods when their detection capabilities are considered.…”

“…Figure 12 shows PR, RR, and DA values for the proposed method and other [1][2][3][4] for test videos given in Table 2. While the methods in [1][2][3][4] have approximately 0.3, 0.82, 0.94, and 0.98 PRs, respectively, the proposed method gives PR = 1. Moreover, recall rates and detection accuracy results indicate the superiority of the work compared to similar works.…”

“…Moreover, recall rates and detection accuracy results indicate the superiority of the work compared to similar works. Method Execution time (s/frame) Wang et al [1] 294.67 Lin et al [2] 140.60 Yang et al [3] 0.41 Guzin et al [4] 0.063 Proposed method 0.07…”

“…Furthermore, the method applies the shrinking/expanding approach on the forgery line to eliminate false matches. Experimental results show that the proposed method gives better detection accuracy with improved run time when compared to similar works in the literature [1][2][3][4]. The paper is organized as follows.…”