Important Note: This article is part of a series in which TechReport.us discusses the theory of Video Stream Matching.
6.1 Conclusion
This research report and software implementation provide a comprehensive understanding of video comparison techniques and the requirements of a system like VSM. The findings emphasize how to effectively handle video frames, employing various mean frame extraction techniques to discern mean frames. Four key features—Histogram, Slope, Edge, and Wavelets—are extracted within the VSM, contributing to informed decision-making. The application of the “Dual Classification Technique” and other methodologies enhances the classification process.
Classification is a critical component of this system that requires improvement, particularly in mean frames extraction. The initial classification step relies on a statistical approach, while the subsequent step offers possibilities of a statistical approach, pixel-by-pixel comparison, or the subtraction of two images.
6.2 Future Work
The current system requires further testing with additional video datasets, especially those containing highly dynamic shots. Although different types of video shots were used in the prototype system test, the scene structure in these shots was generally uniform and slowly varying. Investigation into motion-based features, particularly for highly dynamic video shots, is necessary. The current work focuses on similarity analysis using discrete video shots, but future efforts will extend the system to perform similarity analysis using entire video sequences, eliminating the need to process mean frames separately.
The future lies in videos, as traditional formats such as hyper text and printed text gradually diminish. History has seen a shift from text to audio, then to hyper text due to the internet. Presently, newspapers are publishing less on paper, moving towards digital platforms. A breakthrough in information consumption is emerging with mobile communication, especially through SMS news alerts, which surpass the speed of traditional internet, newspapers, and TV. With TV news channels gaining prominence, there is a growing need to manage their video databases, and VSM emerges as a solution to address these challenges. The core future work in this application pertains to database design.
