Eptacam MPEG-7 Network Camera

The Eptacam MPEG-7 network camera extends the capabilities of conventional cameras by analyzing its scene in real-time to generate annotated video that provides an automatic interpretation of the video content in the way humans perceive visual information.

A video stream for conventional display rendering is encoded in JPEG and annotations that contain video interpretations are encoded according to the MPEG-7 standard. The video stream can also be automatically altered by the Eptacam to hide people's identities and to protect privacy.

The annotated video provided by the MPEG-7 camera is used by the receiving end to extract semantic information of the scene and for a context-aware stream that allows for adaptive back-ends that can analyze content changes in the stream for further processing at a higher level.

Use in Video Surveillance

The Eptacam has a very wide range of applications in video surveillance as the XML-based MPEG-7 encoding is used for the scene description. Meaningful objects in the scene are automatically extracted and their physical properties such as position, speed, shape, color are available as XML tags at the receiving end.

Preventive video surveillance tasks like objects left or removed, trespassing, tail-gating, or people counting can be notably facilitated and automated by using the object-based textual description generated by the MPEG-7 Eptacam. Events are automatically detected by processing relevant object parameters in the back-end, such as a person position to determine trespassing.

Moving the key processing for video analysis to the edge of the network represents a partitioning of the system that creates unlimited scalability and highly cost-effective implementation.

Adaptive Back-end

When critical events are detected a back-end viewing station for the MPEG-7 camera responds by analyzing image content. This approach completely reverses the alert generation logic, providing the back-end with the context information needed to extract high-level information from the video stream. For instance, upon detecting changes in the content that signal an object left behind or other potential security breaches, a viewing system could be programmed to automatically and smoothly change from low-resolution to high-resolution, or create bigger images, and/or issue audible alerts.

Demos