Development of a Video Monitoring and Watermarking Backend in Python Using FFmpeg-Generated Streams
Development of a Video Monitoring and Watermarking Backend in Python Using FFmpeg-Generated Streams
Iman Rahmanian Koshkaki1 Mostafa Abdollahian Dehkordi2
1) B.Sc. Student in Computer Engineering, National Skills University of Shahrekord, Iran Email:
2) Department of Electrical and Computer Engineering, National Skills University of Tehran, Iran Email:
IOI Code
:
AAHD-ABHFCK
Publication :
National Conference on Information Technology, Nanotechnology, Artificial Intelligence and Technological Futures Studies(itnaf.ir)
Abstract :
The rapid proliferation of video surveillance systems in Internet of Things (IoT) networks has introduced fundamental challenges, such as high cloud network latency and vulnerability to visual layer attacks (e.g., deepfakes). Traditional network encryption methods only protect the transmission medium and fail to guarantee the authenticity of pixel content post-decryption. In this research, a novel edge computing-based architecture is proposed for the real-time authentication of IP camera video streams. By offloading the processing load to the proximity of data generation sources, this system leverages the FFmpeg multimedia framework for the optimal extraction of video streams, alongside the OpenCV library in a Python environment to apply spatial domain watermarking algorithms. Laboratory evaluation results demonstrate that the proposed architecture is capable of injecting security marks into video streams with negligible processing latency (approximately 2 milliseconds for 720p resolution) while fully preserving the base frame rate (30 frames per second). Furthermore, CPU utilization for high-quality videos was consistently controlled and low, proving the efficiency of this lightweight approach for deployment in local gateways. This operational solution not only eliminates the communication bottlenecks inherent in centralized systems but also ensures the security, integrity, and immutability of video evidence before the data leaves the local environment.
Keywords : video surveillance systems edge computing real time authentication digital watermarking visual layer processing internet of things (iot)