Principles of operation of a CCD

Here we look at the basic principles - How does a CCD work?

What is a CCD?

In a digital camera the traditional photographic film is replaced by a Charge Coupled Device (CCD).

A CCD is a mosaic of  tiny light sensitive detectors called pixels or 'photosites'. The pixels are arranged as a flat rectangular surface onto which an image is projected using a camera or telescope lens. Each pixel accumulates an electrical charge depending on the amount of light falling upon it.

When an image is 'captured' the electrical charge from each pixel is measured and converted to a number (digitised) by the electronic circuits within the camera. These numbers are transmitted to a computer (immediately or at some later time) where they are used to control the brightness of points on the computer screen (screen pixels), thus reproducing the original image projected onto the CCD.

Sets of numbers representing a complete image are stored in the computer as 'image files'. There are many powerful software programs available to process these files and enhance the image by adjusting the contrast, colour balance etc.

Signal and Noise

The efficiency with which CCD pixels can capture faint images is much better than traditional photography,  but there are problems that need to be understood. Most of these are relate to 'noise'.

The variation in the brightness of pixels across the image shows the 'picture' to our eyes. The greater the variation, the greater the contrast. However, variation can come from:

• Signal: Pixel variation due to the features of the object being photographed.
• Noise:  Pixel variation due to other unwanted causes.

For a good image we need to have a good signal-to-noise ratio. If there is too much noise we will have an image with poor contrast, speckles or other unpleasant features. (See this example of a noisy image of C/2004 Q4)

A CCD will introduce noise through bias, dark current, quantum noise and inhomogeneities, each of which is described below. Modern, cooled cameras developed for astrophotography will suffer relatively little from these problems, but for long exposures they can still degrade image quality. Webcams and ordinary digital cameras are generally very 'noisy'.

Bias (offset)

Each CCD pixel will have a certain minimum electrical charge even if the exposure is of zero duration. This means that pixel values will always be greater than zero even for the shortest exposures.  This is called 'bias' or 'offset'.

Dark Current

Each CCD pixel will accumulate additional charge during the exposure even if there is no light falling on it. This means that pixel values will be higher than they should be by an amount that depends on the length of the exposure. This is 'dark current'.

The effect of Bias and Dark Current is that all parts of the image are a bit brighter than they should be. The process of Calibration aims to remove this excess brightness by 'Dark Frame Subtraction'.

Flat Field Inhomogeneities

The sensitivity of pixels may vary across the surface of the CCD. Some pixels may be 'dead' (giving no response) or 'hot' (always giving the maximum value). Bias or dark current may vary from pixel to pixel. Light or heat leakage inside the camera can cause bright areas or spots. All these effects contribute to CCD flat field inhomogeneities.

In addition, (and as for traditional photography), there may defects in the optical system such as vignetting, internal reflections and 'do-nut' shaped shadows from dust particles on the CCD window.

These defects can cause unwanted light/dark areas or points in the image. The process of Calibration aims to minimise these defects by 'Flat Fielding'.

Quantum Noise

There are random variations in the charge of each pixel due to the physical processing going on within them. This means that each pixel may not return the same value each time it is exposed to the same amount of light. These random variations apply to bias, dark current and flat field inhomogeneities as well as the desired signal.

The effect is that an image has a grainy look and fine detail may be lost. There are a number of approaches to quantum  noise reduction discussed later.