April 2026 Newsletter

Image Sensor Noise model for Image System Simulation

Presents an image sensor noise model based on PTC, combining dark noise, shot noise, and PRNU. Coefficients are optimized for an accurate fit, enabling the prediction of imaging performance, including low-light, and supporting both classic and information-based metrics.

Introduces a dynamic range and low-light measurement method using C₄ information capacity from ISO 12233 slanted edges across exposures. It better reflects real performance and accounts for stray light, improving over SNR-based methods.

A method for calculating NIR bandpass-adjusted Optical Densities for better matching common standard test chart specifications

Presents a camera-matched method for designing NIR test charts by aligning optical densities with system bandpass. Uses spectral measurements and modeling to ensure accurate, application-specific performance across NIR-sensitive imaging systems.

From Centroid to Low-Pass Edge Fitting in ISO 12233 eSFR: Accuracy and Impact on Digital Imaging Information Metrics

Evaluates edge localization in ISO 12233 eSFR, showing centroid methods introduce noise-driven bias and blur, while low-pass filtering improves accuracy. Impacts SFR, information capacity, and SNRi, enabling improved analysis methods.

Validation of Skin Tone Test Charts with Real Human Data

Develops and validates skin tone test charts spanning the Monk Scale using surveys, controlled photography, and measurements. Aligns camera color and exposure with real skin, improving fairness and accuracy across applications.

Imatest will be at Booth #907 of the Premier Conference for Innovators Incorporating Computer Vision and Physical AI in Products.

Visit our exhibit on May 11 & 12. There will be a “Deep Dive” at the Qualcomm AI Hub on May 13.

Imatest’s Simatest module, designed to simulate complete camera systems, can be used with images you generate from lens design software such as Keysight’s Code V or Synopsys’ Zemax OpticStudio.

These images will include optical aberrations, but will not include sensor noise, sharpening, or other effects that an ISP might apply to an image. Input your generated images into Simatest to produce results for image quality analysis.

This paper proposes a hybrid quantum–classical denoising model (QDAE) for retinal fundus images used in diabetic retinopathy detection. By combining convolutional encoding with quantum circuits, it improves noise reduction and detail preservation. Results show superior performance over classical methods, supporting its use as a preprocessing step.

Follow Us