<![CDATA[The advancement of digital multimedia technology has made video manipulation increasingly accessible, making video authentication in digital forensics more critical than ever. This study analyzes interframe Video Forgery patterns using Lucas-Kanade and Farneback Optical Flow methods to detect frame insertion, frame deletion, and frame duplication across 10 manipulation videos sourced from the Kaggle platform. The research stages include frame extraction, grayscale conversion, Optical Flow estimation, and motion magnitude pattern analysis. Each manipulation type produces distinctive patterns, frame deletion generates a single magnitude spike, frame insertion produces two spikes at the entry and exit points of the inserted segment, and frame duplication yields a repeating pattern between the original and duplicated sequences. Both methods successfully detected 9 out of 10 datasets with different failure cases, demonstrating that their complementary use provides more comprehensive detection coverage for temporal forensic analysis of manipulated videos.]]>
Copyrights © 2026
