(This video is from EngineerGuy series #4)
Would you like to translate the captions for this video? Visit http://www.engineerguy.com/translate
Photographic technology has changed a lot in the last century. Digital photography became cheap and convenient because of the CCD: The charge coupled device inside this camera. In the past we’d just place a piece of film here behind the lens, but now we have a purely electronic imager to do all the work. Here it is.
Here it is, just a shiny slab, yet its details are fascinating. This CCD captures the image, and then transfers it to the camera’s memory system to record it as electronic data. When exposed to light, different sections of this CCD build up electric charges proportional to the light’s intensity. We can then measure that charge, and know precisely how bright that section of the image should be. If you enlarge a picture on your computer you can easily see the tiny picture elements, called “pixels.” Each one corresponds to a single section of the CCD.
This is the essence of the digital camera: Several million of these sections of photosensitive silicon in a grid capturing an image. Now, the key to understanding how digital photography became cheap and ubiquitous lies in the ingenious way that the CCD transfers the image into the camera’s memory. The easiest and most straightforward way would be to use wires to connect each pixel in an x-y grid, but that presents a problem.
Here’s what goes wrong. After exposing the grid to light, we can use these wires to read the information pixel by pixel, measuring the charge on each section of the grid. This makes sense in principle, except all of the pixels and their electronic components leak a small bit of charge that distorts the charge coming from each pixel and leads to striations and patterns in the image. This distortion, called capacitive coupling, increases as the number of pixels increases. A CCD solves this problem in a very simple way: the pixels have no wires attached to them!
A CCD is made from a slab of silicon. To make each pixel within the slab, engineers create insulating sections called channel stops. These divide the slab into rows. The surface is covered with a thin layer of insulating silicon dioxide, and then perpendicular to the channel stops, engineers deposit thin strips of metal, typically aluminum. Each pixel, then, is one section bound by channel stops and aluminum.
So now we have our grid of pixels! Recall that when light strikes the whole array, the silicon pixels build up charge proportional to the intensity of the light striking them. We then have a captured image stored as charge within each pixel. The great innovation of the CCD was how it moved the image from the silicon array to the camera’s memory system without using external wires that would distort the image. The CCD shifts the charges from row to row - without wires! - until they reach the bottom where a read out register transfers the charge to the camera’s memory with very little distortion, the camera then counts the charges and constructs the image using that data.
Now, one last important question: How do we get color? A CCD only detects total light intensity, which is useful for producing a black and white image, but for a color photograph we need to separate the entering light into red, green and blue. The most obvious solution is to filter the light into those three colors, using three separate CCDs to capture red blue and green information and combine them into a full color image.
However, engineers have created a cheaper solution: Instead of using three separate CCDs, they use a little math so they only need one. In this consumer camera they cover that single CCD with a filter with red, green and blue pixel sized sections. This creates an image coming out of the CCD that’s a mosaic of these three colors. The camera applies an algorithm to estimate the correct colors for each pixel.
For example, if a green filter covered a pixel we would need to estimate the red and blue components of that pixel. To do this they’d use the adjacent pixels and average the color intensity for the pixel in question. It works because the image’s significant details are much larger than each pixel. This sounds implausible, but you’ve seen the results yourself!
I’m Bill Hammack, the Engineerguy.
This video is based on a chapter in the book Eight Amazing Engineering Stories. The chapters features more information about this subject. Learn more about the book at the address below.