Shopping? Check out our latest product comparisons

Memory chips could lead the way to gigapixel cameras

By

October 14, 2009

A Dutch research team used memory chips to create affordable high-resolution image sensors...

A Dutch research team used memory chips to create affordable high-resolution image sensors for digital cameras. (Photo: Wikimedia Commons, released under cc-by)

Image sensors embedded in digital cameras are expensive, and issues with their circuitry limit the quality and resolution in the pictures they produce. Now a research group from the Netherlands believes a cheaper solution could be right before our eyes - the team's "gigavision" technique exploits the high light sensitivity of memory chips to produce inexpensive gigapixel sensors that perform very well, especially in extreme lighting conditions.

Today's image sensors in digital cameras are based on CCD or CMOS technology, which is effective but relatively complex and not very energy-efficient. With CMOS sensors, as light hits the objective its intensity is translated to an analog voltage for each pixel; the voltage is then transferred to the edge of the chip, where it's converted to a greyscale value between 0 and 255 by an analog-to-digital converter (ADC) in a process that often compromises image quality.

Memory chips store digital information as an electrical charge, which needs to be very small to make read and write operations faster and store more data on the same surface area. However, this makes them very sensitive to external sources of noise such as light, which can easily alter the bits' values as photons hit the transistors in the chip.

A team led by Edoardo Charbon of the Technical University of Delft, Netherlands, decided to exploit this phenomenon by conceptually "removing" the black plastic packages that protect memory chips from interfering photons. The team mapped the light hitting the camera's objective directly to the chip's memory cells, which act as a myriad of miniaturized digital sensors.

The main advantage with this approach is the better resolution with reduced complexity, because it can achieve imaging with 100 times as many pixels on a chip of the same area, reducing the cost in the meantime. Since most digital cameras can now take pictures at 10 megapixels, this means affordable gigapixel cameras could be easily manufactured.

This could pave the way for inexpensive digital cameras in cell phones or other devices that can take better pictures especially in conditions of very dim or very bright light, two areas in which digital cameras tend to struggle.

There is however one catch: each bit can only be set to either "0" or "1," as opposed to a the traditional greyscale sensor that is translated to one of 256 possible levels. This means that sensing the right level of gray accurately is trickier and requires an algorithm to consider not just one, but rather a few adjacent bits at a time.

The team is still working on developing an efficient algorithm for greyscale detection, and is hoping to have a first version of a gigavision memory chip manufactured late this year and a definitive version early next year.

Via New Scientist.

About the Author
Dario Borghino Dario studied software engineering at the Polytechnic University of Turin. When he isn't writing for Gizmag he is usually traveling the world on a whim, working on an AI-guided automated trading system, or chasing his dream to become the next European thumbwrestling champion.   All articles by Dario Borghino
Tags
2 Comments

This technology can revolutionize the quality of security cameras and reduce the cost. Huge innovation.

Facebook User
15th October, 2009 @ 07:22 am PDT

This one was done by Byte magzine some 20 years ago. In those days it was 16 or 64 megabit dynamic RAM chips, but it was the same idea.

To resolve only 256 levels of grey scale would take 256 bits. Typical cameras resolve 4096 level (my DSLR 16,000). There goes your resolution.

TANSTAAFL - There Ain't No Such Thing As A Free Lunch.

splatman
21st October, 2009 @ 03:09 am PDT
Post a Comment

Login with your gizmag account:

Or Login with Facebook:


Related Articles
Looking for something? Search our 28,257 articles