研究目的
To develop a mobile cloud-based video surveillance framework called SmartMobiCam that leverages smartphone cameras and cloud services for smart city applications, aiming to detect, deter, and share malicious activities for public safety.
研究成果
The SmartMobiCam application effectively leverages smartphone cameras and cloud computing for smart city video surveillance, outperforming traditional systems. Performance analysis indicates variable waiting times due to network and cloud factors. Future work includes developing adaptive resource provisioning and adding automatic context tagging for enhanced content review.
研究不足
The performance is affected by network type, signal strength, internet connection, and cloud resource provisioning, making precise delay estimation difficult. The system relies on user participation and may have scalability issues with high numbers of requests.
1：Experimental Design and Method Selection:
The study designed an Android application (SmartMobiCam) that captures videos using smartphone cameras, uploads them to a cloud service (Amazon EC2) for processing, and distributes enhanced videos to subscribers. Python is used for video processing, and Ubuntu is the working environment. The design includes video enhancement (brightness/hue/saturation adjustment) and restoration (Non-local Means Denoising algorithm).
2：Sample Selection and Data Sources:
The application was tested with videos captured by users in smart city scenarios, such as traffic updates and accidents. Specific hardware used includes a Sony Xperia M phone with a megapixel camera, 1GHz dual-core processor, 1GB RAM, and up to 32GB storage.
3：List of Experimental Equipment and Materials:
Smartphone (Sony Xperia M), Amazon EC2 cloud service, WiFi connection with upload speed ≈1500kbps and download speed ≈3500kbps, Eclipse IDE for app development, Android AVD for testing, Python for cloud processing, Ubuntu OS.
4：Experimental Procedures and Operational Workflow:
Users open the app, capture a video, choose to upload it to the cloud. The video is processed in the cloud using Python, enhanced and restored, then sent back to the user or shared via social media. Performance is evaluated based on average waiting time (Ttot), which includes processing time, transmission times, and network latency.
5：Data Analysis Methods:
The mean waiting time for 10 transmissions was calculated to analyze performance. Equations (1), (2), and (3) from the paper define Ttot, Tproc, Tdown, and Tup, considering network conditions and cloud processing delays.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
