03 Oct 2003
It may be hidden deep inside your computer, but a good frame grabber is crucial if you want to perform high-quality scientific imaging. Oliver Graydon offers the low-down on this vital piece of electronics that enables high-speed image transfer from cameras.
From Opto & Laser Europe October 2003
Without this hidden piece of electronics it is not possible to transfer images from the camera - or any other equipment with a video output - to a computer. Today, most frame grabbers suitable for scientific imaging come in the form of printed circuit boards that simply plug into a vacant PCI-bus slot in a computer.
So what do frame grabbers do? Their main function is to digitize a video image into an array of digital data-points (pixels) that can be stored, processed and displayed on a computer. A second useful feature is that they often come with a series of outputs for powering and controlling a camera.
Frame grabbers that are designed to operate with cameras with a digital interface (such as CameraLink or LVDS) are not discussed here, as these do not convert a raw analogue video image into digital data. If you're lucky enough to have a camera with a digital interface then do use it as you will benefit from higher frame-rates and better pixel quality.
Quality counts All of the frame grabber makers that Opto & Laser Europe spoke to wanted to emphasize the distinction between products that are designed for multimedia applications (such as transferring images from a webcam onto the Internet) and those suitable for more demanding scientific or industrial tasks.
Although the former type can be purchased from a computer outlet for less than £50 (€72), they do not offer the low-noise performance that is necessary for high-quality imaging. Nor do they come with the software needed to perform complex camera control or sophisticated image manipulation. The price for a typical frame grabber for scientific applications starts at a few hundred pounds and goes up to a few thousand.
So what do you need to bear in mind when buying a frame grabber? Many of your decisions will be dictated by the type of camera you have chosen. Check whether your camera outputs colour or monochrome images, and find out whether the video signal is a standard composite signal (RS-170, NTSC, CCIR, PAL, SECAM) or a non-interlaced format such as a progressive scan image.
Features and specifications Once you are satisfied that your camera and frame grabber are compatible, it's time to start thinking about other features that you may want. To simplify this task Opto & Laser Europe has compiled a list of the criteria you will need to consider, along with a short description of each item.
Camera control outputs Some frame grabbers can provide a 12 V output for powering a camera. This means that an external power supply is no longer needed. Other outputs (often TTL digital signals) can be used to trigger the camera and control its exposure time.
Circuit noise Circuit noise is a measure of how well the frame grabber performs the digitization process - and the lower the value the better. Noise is usually specified in units known as LSB (least significant bit).
Comprehensive software support Find out what kind of software and programming languages the frame grabber supports. For example, does it support third-party packages, and does it come with all the software drivers that you need? This is particularly important if you plan to perform a large amount of image processing.
Digital signal processors Some frame grabbers incorporate digital signal processor (DSP) chips. These allow the frame grabber to perform very fast preprocessing of the image, such as image filtering for noise reduction, before an image is transferred to the computer. The idea is that this reduces demand on the host computer and is a fast and efficient way of converting the images into their final form. It can be useful if you need to carry out lots of heavy-duty image manipulation, but if you simply wish to transfer the raw images from the camera it is not necessary.
Gain and offset controls A gain control makes it possible to adjust the brightness of your image so that its pixels occupy an optimum range of grey scales (maximizing its dynamic range). This is a useful way to enhance the contrast and quality of an image that has been imaged in poor lighting conditions.
An offset control allows the absolute brightness of all the pixels to be shifted up or down by the same amount to compensate for an image that is too dark or too bright.
Lookup table After a frame grabber has digitized an image, the brightness values for each pixel are stored in a giant "lookup table" (LUT). Two types of LUT exist: input LUTs allow the user to change the value of a pixel prior to it being stored or displayed, allowing very fast preprocessing of an image; whereas output, or palette-matching, LUTs allow the user to change the displayed image. This has the advantage that the stored image is left in its original form.
Multiple inputs If you want to connect more than one camera to the computer, make sure that the frame grabber provides multiple inputs. Most frame grabbers have a built-in multiplexer that allows four or more cameras to be connected.
Onboard video memory Many frame grabbers come with a fairly large onboard memory. This acts as a buffer for the temporary storage of images before they are transferred to the computer. Today, the high data-transfer rate of PCI buses (up to 132 Mbytes/s for a 33 MHz bus and 533 Mbytes/s for a 66 MHz bus) means that a large memory is not usually necessary, but there are two exceptions: first, if you are dealing with images that have a very high resolution and large colour bit-depth and are thus exceptionally memory-hungry; and second, if the PCI bus in the computer is being shared with several pieces of hardware which are fighting for its use.
Synchronization timing and pixel jitter Frame grabbers need to synchronize their onboard clocks to the timing signal of the incoming video data. This is achieved using either a phase-locked-loop (PLL) circuit or a crystal-controlled digital clock synchronizer (DCS). DCS circuits tend to offer better performance than PLL circuits, with lower pixel jitter and faster resynchronization, but they are more expensive.
Pixel jitter is a measure of how precisely a frame grabber can sample the input image. Poor performance means that pixels are not properly positioned in the image. Ideally this value should be as low as possible; typically it is a few nanoseconds.