In this tutorial, we present a guide to the multidisciplinary science of building the eyes of a vision system. CMOS image sensors have been instrumental in lowering the barrier for embedding vision into systems. Their high degree of integration allows photons to be converted into bits with minimal support circuitry. Simple protocols and interfaces mean that companies can design camera-based systems with comparatively little specialist expertise. To produce high-quality output, the image sensor and optics must be carefully co-optimized to fit the application. To assist with component selection and help avoid common pitfalls, we will describe the key parameters and provide a practical guide to selecting both sensor and optics for a camera. We’ll also provide an introduction to other hardware considerations and to correcting optical aberrations in the image processing pipeline.
Building the Eyes of a Vision System: From Photons to Bits
Session Speakers
-
Jon Stern
Director of Optical Systems, GoPro
Jon Stern has over 25 years of experience in imaging, digital cameras, and image sensors. He graduated with a Ph.D. in 3D chip stacking from the University of Sheffield in the United Kingdom in 1995, where an encounter with an early CMOS image sensor led to a career in imaging. This was never intended, despite growing up in a household full of cameras. Since then he has worked in equal measure using CMOS image sensors while at camera companies, and in supporting others in integrating them while at sensor manufacturers. At ON Semiconductor (formerly Aptina), he led the camera module engineering team, while at AltaSens he was head of applications engineering. Currently, he is Director of Optical Systems at GoPro, where he leads a team that develops the imaging front ends for their popular action cameras.
