An Adaptive Steganography Approach for Live Video Streams Based on Edge Analysis and Frame Variability

With the rapid evolution of digital video communication, the field of steganography has gained more attention. Video steganography has vital advantage over other steganographic techniques due to its utilization of video as the cover medium for data hiding. In this paper the advantages of using live video as a cover medium for steganography is explored. This paper presents an adaptive steganography which uses live video as cover medium. The proposed method uses content analysis and frame heterogeneity to dynamically select several steganographic techniques. Live frames are processed by initially applying Canny edge detection, yielding an edge map. Edge density is then calculated, enabling the identification of smooth frames. Depending on the calculated edge density, the algorithm chooses between Canny and Sobel edge detection techniques to adapt to diverse content complexities. Additionally, each frame's standard deviation is calculated to determine variation in content of the frame. If the computed standard deviation surpasses a predetermined threshold, LSB is selected as embedding technique. Otherwise, the Discrete Cosine Transform (DCT) is favored for its computational efficiency. The findings suggest that our methodology is apt for real-time video applications, as indicated by the time taken parameter. Additionally, the proposed method's superiority is substantiated by the MSE and PSNR metrics.