A Selective Change Driven System for High-Speed Motion Analysis.

Area of Science:

• Computer Vision
• Robotics
• Sensor Technology

Background:

• Vision-based sensing algorithms are computationally intensive due to large data volumes.
• Conventional sensors often provide redundant data, straining processing capabilities.
• Selective Change Driven (SCD) sensing offers a more efficient data acquisition strategy.

Purpose of the Study:

• To develop and implement a novel Selective Change Driven (SCD) processing architecture for high-speed motion analysis.
• To leverage pixel-based processing instead of full frames for enhanced computational efficiency.
• To demonstrate the feasibility of SCD combined with hardware processing for real-time applications.

Main Methods:

• Developed a new SCD processing architecture implemented on a Field Programmable Gate-Array (FPGA).
• Utilized a 64x64 CMOS SCD sensor for data acquisition.
• Implemented a system that processes individual pixels, controlling the data stream for real-time object distance calculation.

Main Results:

• The SCD system achieves object distance calculation in just 1.7 microseconds per pixel.
• This pixel-based processing approach bypasses the need for extremely high frame rates (e.g., 500 Kilo Frames Per Second) required by conventional methods.
• Demonstrated the system's effectiveness in high-speed motion analysis.

Conclusions:

• The SCD approach, when integrated with hardware processing, significantly enhances the efficiency of vision-based sensing.
• This system offers a practical and feasible solution for high-speed motion analysis, overcoming limitations of traditional frame-based cameras.
• The developed architecture shows the substantial potential of SCD sensing for computationally demanding vision tasks.