Optical flow in a smart sensor based on hybrid analog-digital architecture

Area of Science:

• Computer Vision
• Embedded Systems Engineering
• Robotics

Background:

• Optical flow estimation is crucial for real-time motion analysis in various scientific domains.
• Existing methods often face computational challenges, limiting real-time performance.
• Developing efficient, integrated solutions is essential for practical applications.

Purpose of the Study:

• To design and implement a novel motion sensor for optical flow computation.
• To create a System-on-a-Chip (SoC) integrating focal plane processing and a digital processor.
• To achieve real-time performance with reduced power consumption.

Main Methods:

• Developed a simplified optical flow model for efficient computation.
• Implemented the model on an SoC combining analog focal-plane and digital NIOS II processors.
• Employed co-design techniques for optimizing the integrated architecture.

Main Results:

• The developed system achieves real-time optical flow estimation.
• The architecture demonstrates high efficiency by leveraging focal plane pre-processing.
• Performance and accuracy were evaluated against existing literature approaches.

Conclusions:

• The study successfully created a low-cost smart sensor for optical flow.
• The integrated SoC architecture provides a viable solution for real-time motion sensing.
• The system's efficiency and reduced power consumption enable diverse application domains.