During my internship at Dallmeier electronic, I worked on two groundbreaking projects in the field of video surveillance technology. The focus was on the mobile availability of livestreams on the one hand, and the intelligent automation of camera systems on the other.

Project 1: iPhone Video Surveillance

At a time when the iPhone (and iPod Touch) were revolutionizing the market, we investigated the suitability of these devices for security applications. The goal was to enable security personnel to access camera streams via mobile.

Project 2: Automatic PTZ-Dome Tracking

The second project was a research study on the possibilities of automatic object tracking with pan-tilt-zoom (PTZ) cameras.

Experimental Setup

A static wide-angle system provides overview information, while a PTZ dome system detects and tracks the recognized objects in detail.

Tested Algorithms (OpenCV)

  1. Optical Flow (Lucas-Kanade): Promising in theory, but problematic in practice with moving cameras. Motion blur made feature extraction significantly more difficult.
  2. Background Modeling & Difference Image: By combining motion detection (difference between consecutive frames) and an iteratively updated background model, objects could be reliably detected even if they came to a brief standstill.
  3. Stop & Go Method: To eliminate motion blur during analysis, an iterative method was tested where the camera moves step-by-step and analysis is performed only when stationary.

Conclusion and Technologies

The internship provided deep insights into professional image processing under Linux.