EI2025 Short Course: Information metrics for optimizing Machine Vision systems

Description: A new set of metrics based on information theory promises to be superior to traditional MTF (SFR or sharpness) and noise for predicting machine vision system performance. We will introduce the new information metrics, which include Noise Equivalent Quanta (NEQ), camera information capacity, Ideal Observer SNR (SNRi – for the quality of object detection), and Edge location standard deviation (Edge σ – for the quality of edge location). We will cover the background of the new measurements, why they are more directly related to object and edge detection than traditional measurements, how to conveniently obtain them (primarily from standard slanted edge test patterns), how to interpret them, and how to design matched filters for optimum system performance.

Level: Intermediate

Length: 2 hours

Instructor: Norman Koren, founder & CTO Imatest LLC

Prerequisites: Knowledge of basic image quality measurement concepts, especially MTF (SFR) and noise

Benefits:

• the history and mathematics of information theory in imaging,
• definitions and interpretations of the information metrics, which include camera information capacity, Noise Equivalent Quanta, Ideal Observer SNR (SNRi – a metric for the quality of object detection), and Edge location standard deviation (Edge σ – a metric for the quality of edge location).
• How to conveniently obtain the information metrics
• the effect of common types of image processing on metrics, including uniform sharpening and lowpass filtering (for noise-reduction), as well as the nonuniform bilateral filtering found in most camera JPEG images,
• design of matched filters to optimize SNRi and Edge σ,
• progress in correlating the new metrics with machine vision performance and developing ISO 23654

Intended Audience: Engineers who design and analyze cameras and imaging systems for automotive, medical, security applications, and more

Format: Lecture Primarily lecture, but about 1/4 to 1/3 of the time will be demonstrations

Key Words: MTF, information capacity, Artificial intelligence, machine vision.

Registration is open – Click Here to register

Electronic Imaging 2025, held February 2–6 at the Hyatt Regency San Francisco Airport, offers a vibrant platform for industry and academia to advance imaging technologies. The event features plenary talks, keynotes, technical sessions, networking opportunities, an industry exhibit, and a short course program, covering all aspects of electronic imaging in a dynamic and collaborative environment.

Learn more about Electronic Imaging 2025

If you have any questions, please contact info-metrics@imatest.com.