Thales Optronics 03 (2009-13)

Title: Embedded processing: automatic image and scene analysis

Research Engineer: Calum Blair
Sponsor: Thales Optronics
Academic Supervision: Dr Neil Robertson, Heriot-Watt University

This project is in the technical area of digital image processing and computer vision. The main focus of the project is hardware implementation and not algorithm development. Algorithm development is being explored in a concurrent EngD with the VISP Laboratory at Heriot-Watt University and Thales Optronics.

The aim of this project is to research and develop techniques for easily implementing novel automatic image analysis algorithms for scene understanding. The output of the project will be a real-time technology demonstrator. This will be built using appropriate hardware which may include SoC, GPU and FPGA.

Automatic image and scene analysis algorithms involve automatically decomposing a video stream into “threat areas”. Specific processing and decision-making is then independently applied to each area depending on the nature of the threat. These context-driven scene analysis algorithms automatically select the appropriate sensor and processing algorithms for each threat area. This project will research the appropriate software and hardware required to implement the algorithms in real-time. The project will explore the intelligent allocation of the individual algorithms to software and firmware with an emphasis on the real-time implementation in hardware.