Optimizing camera system development: a robust simulation tool for advanced vision applications

Traditional image quality metrics often fail for fine-tuned applications. They require advanced simulation software to meet specific system requirements while minimizing cost and development time. Although existing tools are complex and do not facilitate efficient communication between stakeholders, we developed a comprehensive simulation tool for end-to-end vision system modeling. It supports various complexity for all components (optics, sensors, ISP, and computer vision tasks), enabling users to simulate a vision system from the early design phase. We will demonstrate its accuracy by comparing it to a commercially available drone camera module and show how manufacturing tolerances help limit discrepancies. In addition, we will present its use in translating application requirements into system specifications, testing standard components, and generating data to build robust models for machine vision.