Algolux has enhanced and updated its CRISP-ML software platform that automates the manually-intensive task of camera image quality (IQ) tuning through machine learning and objective IQ metrics (KPIs). The latest release promises to significantly improve camera team productivity via an improved tuning methodology for increased usability and flexibility to achieve IQ goals.
Updated features include:
- Multi-objective “ground truth” lab charts are generated and displayed on calibrated high definition monitors to provide a more flexible lab environment than just using current physical transmissive or reflective charts (Figure 1).
- Charts quickly tailored to accommodate tuning against different image quality KPI sets or even custom KPIs.
- KPI sets now apply an intuitive threshold approach to guide the CRISP-ML solvers to optimize the highest priority metrics.
- Users can quickly define different tuning campaigns to best meet their requirements.
- Enhanced image quality tuning and optimization analysis from system to component level
- Deep analysis capabilities now allow users to traverse each parameter iteration to determine optimizer convergence and quickly filter through the tens of thousands of iterations to prioritize top results for further analysis and visual inspection of tuned images (see figures 2 and 3 below).
- Sensitivity analysis enables exploration of parameter, KPI, and image signal processor (ISP) block dependencies, providing insight into their impact to image quality.
- Stability analysis helps determine the most stable parameter setting against KPI targets to help mitigate IQ variances due to affects such as device yield and temperature.
- Accelerated bring-up of new and archived tuning projects
- New database repository architecture stores all tuned images, parameter settings, and KPI scores for each iteration of every tuning run for more efficient analysis and archiving.
- Camera tuning teams have complete access for customer reporting requirements and can quickly restore prior projects to respond to customer requirement changes or to jumpstart new projects, addressing infrastructure challenges faced by teams today.