Understanding CCD Read Noise
Interpret FFTs l
Noise FFTs l
Your Camera l
Create FFTs l
Understanding CCD camera read noise
The goal of every imager using a cooled CCD camera, from amateur astronomers to commercial users and scientfic researchers, is to produce the best images possible. Cooled CCD cameras and digital image processing have revolutionized scientific imaging and astrophotography making it possible for serious amateurs using relatively inexpensive equipment to take pictures rivaling those from large, professional observatories taken just a decade ago.
While CCD cameras have made astrophotography much more accessible to a growing number of astronomers, getting the best possible results from a CCD camera is not without its challenges. Achieving high dynamic range, low noise images from a cooled CCD camera requires a basic understanding of how CCDs work and the different sources of noise that can reduce the quality of your images.
Some types of noise in CCD images are relatively easy to manage using standard image calibration techniques. “Dark current” can be subtracted using a dark frame. Combining multiple images is very effective at averaging out random pixel variations. Flat fields can be used to smooth out pixel non-uniformity and reduce the optical noise in the system. Other types of noise in CCD images are much more challenging to deal with and can make it difficult or impossible to achieve the final image quality you expect.
These pages show you how to identify and analyze the read noise introduced by any CCD camera. Using software that is widely available and easy to use, we provide step-by-step instructions for creating “Read Noise Frames” which isolate the read noise present in a CCD camera. We’ll then show you how to create and interpret Fourier Transforms (FFTs) of those read noise frames to characterize any sources of noise that may be present in your imaging system. This is not intended as a rigorous scientific discussion of CCD noise. We have purposely glossed over some of the more technical issues of measuring different classes of CCD noise in order to concentrate on just the topic of identifying and characterizing the read noise introduced by the camera electronics.
The pages are broken down into these topics.
- Understanding CCD Camera Read Noise
- Measuring CCD Read Noise
- Interpreting FFTs
- Interpreting Read Noise Frame FFTs
- Measuring the Read Noise in Your CCD Camera
- Creating FFTs
You can easily move from page to page by clicking the “continue” link at the bottom of every page or you can jump directly to any section with the links along the top of each page.
CCDs work by converting photons into electrons which are then stored in each pixel. Each pixel can hold a fixed maximum number of electrons, typically from 35,000 to as many as 500,000, depending on the specific model of CCD. While integrating, or exposing, an image, photons strike individual pixels and are converted to electrons and stored in each pixel well. The number of electrons stored in each pixel "well" is proportional to the number of photons that struck that pixel. After an exposure has been completed, the electrons for each pixel are shifted out of the CCD and converted to a number, indicating how dark or light each particular pixel should be, and stored in the image file.
Sources of noise in CCD images
In an ideal world, every photon striking a pixel would be converted into exactly one electron. Then the number of electrons would be precisely counted and converted to a number telling the photographer exactly how much light struck each pixel.
Unfortunately, the process of converting light to pixel values in a CCD image is governed by some fundamental physical laws and other factors that introduce “noise” into an image. Noise is unwanted variations in pixel values that make the image a less than exact representation of the original scene.
Noise in CCD images can manifest itself in multiple ways, including “graininess” in darker background areas, faint horizontal or vertical lines that become visible in low signal areas of the image, blotchy gradients between darker and lighter regions in a nebula, a gradient from dark to light from one corner or side of an image to the other, and especially as low contrast images — the result of a reduced signal to noise ratio.
If each pixel in a CCD can hold 90,000 electrons and the average noise of the system is 30 electrons per pixel, the dynamic range, or Signal to Noise Ratio (SNR) is 90,000 / 30 or 3,000 to 1 (70db). If the average noise can be reduced to just 15 electrons, you have effectively doubled the dynamic range to 76db, or 6,000 to 1 (90,000 / 15). Reducing the noise in CCD images is critical to producing clear, high dynamic range images.
There are several different sources of noise that are introduced during the integration and read out of an image from a CCD. While all noise is problematic because it reduces the effective dynamic range of a CCD camera, some types of noise are worse than others. Somewhat oversimplified, “good noise,” can be reliably reduced with standard image calibration techniques or by combining multiple frames. “Bad noise,” such as electromagnetic interference, is difficult or impossible to reduce through subsequent image processing.
CCD manufacturers measure and report CCD noise as a number of electrons RMS (Root Mean Square). You’ll typically see it presented like this, 15eˉ RMS, meaning that with this CCD, you should expect to see about 15 electrons of noise per pixel. More precisely, 15eˉ RMS is the standard deviation around the mean pixel value.
Below is an overview of the major sources of noise in CCD images and what camera manufacturers or CCD camera users can do to reduce or eliminate them from their images.
CCDs build up “dark current” whether the CCD is being exposed to light or not. Dark current is caused by thermally generated electrons that build up in the pixels of all CCDs. The rate of dark current accumulation depends on the temperature of the CCD but will eventually completely fill every pixel in a CCD. Managing dark current is particularly important for astrophotography because of the long exposures typically required for night sky imaging. The pixels in a CCD are cleared before beginning an exposure, but dark current starts accumulating again immediately.
||The rate of dark current build up can be reduced by a factor of 100 or more by cooling the CCD. The remaining dark current is subtracted from an image using dark frames.
Today’s CCDs are made to exacting standards, but they still are not perfect. Each pixel has a slightly different sensitivity to light, typically within 1% to 2% of the average signal.
||Pixel non-uniformity can be reduced by calibrating an image with a flat-field image. Flat field images are also used to eliminate the effects of vignetting, dust motes and other optical variations in the imaging system.
Shot noise is caused by the random arrival of photons. This is a fundamental trait of light. Since each photon is an independent event, the arrival of any given photon cannot be precisely predicted; instead the probability of its arrival in a given time period is governed by a Poisson distribution. With a large enough sample, a graph plotting the arrival of photons will plot the familiar bell curve.
||Shot noise is most apparent when collecting a relatively small number of photons. It can be reduced by collecting more photons, either with a longer exposure or by combining multiple frames.
CCD Read Noise (On-chip)
There are several on-chip sources of noise that can affect a CCD. CCD manufacturers typically combine all of the on-chip noise sources and express this noise as a number of electrons RMS (e.g. 15eˉ RMS).
||CCD Read Noise is a fundamental trait of CCDs, from the one in an inexpensive webcam to the CCDs in the Hubble Space Telescope. CCD read noise ultimately limits a camera’s signal to noise ratio, but as long as this noise exhibits a Gaussian, or normal distribution, its influence on your images can be reduced by combining frames and standard image processing techniques.
CCD Camera Noise (Off-chip)
A pixel value is read out of a CCD as a tiny voltage, on the order of microvolts per electron. The camera’s electronics then pass that voltage to an Analog to Digital Converter (ADC) to be converted into a digital pixel value. There are numerous opportunities during this process for the camera to introduce additional noise into the CCD image, both uncorrelated, random noise, and the more troublesome periodic noise. It is during this process where the camera’s electronic design determines the quality of the image that the camera can deliver. Two cameras using the exact same CCD can produce final images with drastically different noise characteristics and effective dynamic range.
Before being passed to the ADC the camera must amplify the pixel voltage to a range appropriate for the ADC. All amplifiers introduce noise, whether that amplifier is in a stereo or a CCD camera. Well designed amplifiers exhibit high linearity and introduce minimal noise. Several additional sources of noise, such as Reset noise, Johnson noise, and Flicker (1/f) noise can be greatly reduced using a technique called Correlated Double Sampling (CDS). CDS takes advantage of certain characteristics of CCDs and other electronic devices to precisely subtract out certain types of noise when reading pixel values from a CCD.
||In a well-designed camera, the camera’s electronics add only additional uncorrelated noise as the image is read from the CCD. As with on-chip CCD Read Noise, this random noise can be reduced or eliminated by combining frames.
CCD cameras by their nature contain highly incompatible types of electronic circuitry. The CCD, CCD preamplifier, CDS signal processor and ADC all operate with microvolt level signals and are very sensitive to any sort of electronic interference. Yet, the camera also employs inherently noisy digital electronics to control the camera operation, and usually a switching power supply to provide the internal voltages necessary for operation.
Mitigating the electronic switching noise from the digital electronics and power supply is one of the many challenges in designing a scientific grade CCD camera. This electronic noise can easily interfere with the very precise, low-level CCD signals, introducing significant noise into images. It can enter the circuitry either by radiation (EMI) or by being conducted directly through the electrical connections that tie this sensitive circuitry to the rest of the camera electronics. This type of noise is typically highly correlated in the frequency domain and is often the major contributor of structured noise in a CCD Bias Frame.
||Minimizing the impact of these noise sources in the small, confined space of the CCD camera requires sophisticated mixed-signal design techniques including, careful circuit board layout and isolation, shielding, grounding, signal rise-time control, filtering, and considerate timing.
There are many potential sources of noise in a CCD camera. The read noise inherent in a CCD tends to be random, uncorrelated noise. All cameras add additional noise when reading the image from the CCD and converting it into a digital image. The camera designer’s goal is to add as little additional noise as technically possible, and make sure that no noise is introduced that can’t be reliably reduced using standard image calibration techniques. While combining multiple frames smoothes out random noise, periodic, “bad” noise builds up just like the “signal” from the stars you’re trying to image.
The next section will demonstrate how to measure and characterize all of the noise present in a CCD camera.
Continue to Measuring CCD Read Noise >>