Reconstructing Missing Frame in Video Transmission with Frame Interpolation

This study addresses the issue of missing frame in video transmission, a critical factor affecting Quality of Service (QoS) in video communications. Using spatial interpolation techniques, the research generates replacement frames by analyzing pixels and patterns from surrounding frames, evaluating their accuracy in enhancing visual quality, playback smoothness, and user satisfaction. The FILM Model simulates frame loss scenarios, testing an algorithm designed to recover missing frames and assessing its effectiveness through the Structural Similarity Index (SSIM). Six different videos with four types of transitions were examined, spanning various video types. T-Test analysis across diverse video scenarios indicates that the model performs exceptionally well in Low Lighting and HDR conditions, maintaining high SSIM values and visual quality even at greater frame distances. However, challenges arise with AI-generated and camera phone footage, where SSIM values significantly decline during complex transitions like Cross Dissolve. These differences are statistically significant, highlighting the model's need for improvement in handling diverse video types. The findings underscore the importance of refining the model to better manage dynamic scenarios and complex transitions, enhancing its applicability in broader video production contexts and potentially improving real-time video communication systems.