A self-adaptive parameter optimization algorithm in a real-time parallel image processing system.

Area of Science:

• Computer Vision
• Robotics
• Control Systems Engineering

Background:

• Parallel processed vision servo systems face trade-offs between precision, speed, and robustness.
• Existing load balancing algorithms often require prior knowledge of system load, limiting extensibility.

Purpose of the Study:

• To propose an adaptive load capacity balance strategy (ALBPO) for optimizing servo parameters in parallel vision systems.
• To enhance computing precision and detection ratios while maintaining system robustness and speed.
• To develop a load balancing algorithm with high extensibility, operating without prior knowledge of processor load.

Main Methods:

• Utilizing load capacity (LC) functions to estimate processor load.
• Implementing continuous self-adaptation for load balancing based on fluctuating LC results.
• Selecting target detection and location parameters dynamically based on LC outcomes.
• Evaluating performance against existing load balancing algorithms.

Main Results:

• The ALBPO algorithm demonstrates superior load balance performance.
• Achieved optimization of Quality of Service (QoS) for individual processors.
• Successfully balanced the servo circle, precision, and robustness requirements.
• Showcased high extensibility due to operation under unknown system load conditions.

Conclusions:

• ALBPO effectively addresses the parameter constraints in parallel vision servo systems.
• The proposed strategy offers significant improvements in load balancing and system performance.
• ALBPO provides a scalable and robust solution for optimizing parallel vision servo systems.