Automatic Calibration and Error Correction

Kenton T. Veeder

doi:10.1117/3.1000639.ch6

Translator Disclaimer

Chapter 6:
Automatic Calibration and Error Correction

Book: Digital Converters for Image Sensors

Author(s): Kenton T. Veeder

Published: 2015
https://doi.org/10.1117/3.1000639.ch6

Abstract

Image sensor professionals working regularly with digital image sensors should understand that medium- to high-resolution ADC architectures will need some form of linearization processing in order to generate acceptable linearity for image sensor applications. Linearization is performed on chip for commercial stand-alone ADCs and is transparent to the user. In many academic papers, linearization post-processing is often carried out in software long after conversion; however, this technique is slow and not suitable for any sort of real-time or production-quality image sensor system. Image sensor ADC linearization can be performed on the image sensor chip itself, as is done for ADC stand-alone products, or in the image post-processing electronics operating on the image data in real time, as long as the ADC data stream has been designed to accommodate this post-processing. In the author’s experience, image sensor electronics professionals do not wish to incorporate ADC linearization schemes into the already complex image processing routines running on external processors. This minimizes the number of necessary feedback loops between the image sensor ADCs and external hardware. However, keeping processing off the image-sensing chip can be an important power-saving feature for specialty systems such as infrared focal plane arrays operated at cryogenic temperatures. Again, in the author’s experience, it is worthwhile to linearize the ADC data on the imagesensing chip, even for most cryogenically cooled parts, to enable certain onchip video processing and minimize the complexity of the external video processing engine. The author has found two types of image sensor chips for which the tradeoff is not always obvious. The first is the radiation-tolerant class of parts, where memory can become corrupted by single-event upsets, and external hardened RAM in triplicate provides additional reliability. The second is large digital pixel sensors, where calibration constants are needed for each pixel, and an on-chip secondary memory array might not be worth the circuit area. Whatever type of image-sensing system, it is important for image sensor professionals to understand the complexity of ADC linearization in order to make a good decision about the value of on-chip processing or offchip processing of their ADC data.

Online access to SPIE eBooks is limited to subscribing institutions.

CHAPTER 6
16 PAGES

DOWNLOAD PDF SAVE TO MY LIBRARY