Choosing a Camera


It is important to have the right equipment to achieve success in any project, and machine vision applications are no exception. The choice of a camera can be quite daunting since there are so many different options and manufacturers available. This document discusses the various choices you need to make in the camera selection process. We will focus on all of your options here, but deciding on the features you can afford is left up to you!


Analog or Digital?

One of the most fundamental questions in camera selection is what kind of signal it outputs. The main choice is between analog or digital, but there are also several different permutations within each of those categories. In general, analog cameras are lower cost, but generally have less noise immunity, flexibility in frame rate, and resolution. Unless you have a definite need for high speed, high precision, or operation in a noisy environment, an analog camera will usually be sufficient.

Analog cameras can output either a standard (RS-170, NTSC, CCIR, PAL, and so on) or non-standard signal. Standard cameras are easier to configure, while non-standard cameras provide different characteristics, such as higher resolution or frame rates. If using a non-standard camera, you need to make sure that the frame grabber you choose is configurable to support such a camera.

Digital cameras have some additional choices in the type of signal. Typical data transfer signal types include TTL, RS-422, and Low-Voltage Differential Signal (LVDS, or RS-644). TTL signals are fairly susceptible to attenuation and noise, and often can only be used with extremely short cable runs (a few feet or less in some cases). RS-422 and LVDS are differential signals which are much more robust, with LVDS allowing longer cable runs at lower signal voltages.

See Also:
Anatomy of a Video Signal

Color or Monochrome?

Cameras can provide a color or monochrome signal. The choice of which to use depends on the type of image to be processed; often grayscale intensity information is adequate to extract all necessary information from an image, but in some cases the color information is critical. There are applications which sort identically shaped objects by color, and others where the grayscale contrast is too low to accurately identify objects, but color characteristics are unique; in these cases a color camera is necessary. Color images are usually stored as 32-bit images, and take up significantly more storage space than 8-bit or 16-bit monochrome images.

For analog video, the color information can be added to the monochrome signal to produce a composite color signal. RGB color (red-green-blue) requires three signals -- one for each of the three color planes. RGB color is more accurate, but requires three channels. If the channels share hardware resources, the maximum frame rate achievable will be 1/3 that of a monochrome acquisition on the same frame grabber.

For digital color video, there is not a large selection of cameras on the market, so the choice becomes easier. They are available in either RGB or mosaic color. Mosaic color uses a coding scheme to efficiently represent the color information in the signal.

See Also:
Color Image Representations
Color Matching

Line Scan or Area Scan?

Area scan cameras use a two-dimensional array of sensors to capture an image. The outputs from the sensors are then scanned and transferred to the frame grabber. Line scan cameras have a one-dimensional array of sensors, which capture and output a single horizontal video line at a time.

Line scan cameras are useful for producing a flat image of cylindrical objects, imaging very large objects with good resolution, and producing images of objects moving past a fixed point, such as parts on an assembly line. An area scan camera should be used for any application not specifically requiring a line scan camera.

See Also:
Conveyor Belt Applications

Interlaced or Progressive Scan?

An interlaced camera breaks each image frame into two fields, each containing only odd or even horizontal video lines. The frame grabber or video display device then reassembles, or interlaces, the two fields for display as a single frame. This reduces the perceived display flicker by updating half the display at twice the update rate.

Progressive scan cameras output the entire frame in a single field, so no interlacing is required. This is useful for high-speed applications where precise measurements of motion are needed. Each field is available for display and processing immediately after transfer, instead of being held until the interlacing is complete.

Frame Rate

The standard frame rates for analog video are 30 frames/s for RS-170 or NTSC, and 25 frames/s for CCIR or PAL. Non-standard cameras may produce higher frame rates. The limitation is in how fast the frame grabber can clock in pixels; its pixel clock specification determines the frame rate you can achieve. Cameras with smaller image sizes can achieve higher frame rates at a given pixel clock rate.

Digital cameras can transfer images faster than analog cameras, mainly because they experience less signal degradation during the transfer. This allows for larger images to be acquired at a given frame rate, or higher frame rates at a given image size. The frame grabbers generally support higher pixel clock rates since they don't have to be concerned with A/D conversion issues. The use of multiple-tap cameras allows two, four, or more pixels to be transferred simultaneously, speeding up the frame rate by the same factors.


The term resolution applies to two different concepts with regard to vision systems. Spatial resolution refers to the image size in pixels -- for a given scene, more pixels means higher resolution. The spatial resolution is a fixed characteristic of the camera and cannot be increased by the frame grabber of post-processing techniques. Zooming techniques, for example, merely interpolate between pixels to expand an image without adding any new information to what the camera provided. It is easy to decrease the resolution, however, by simply ignoring part of the data. National Instruments frame grabbers provide for this with a "scaling" feature that instructs the frame grabber to sample the image to return a 1/2, 1/4, 1/8, and so on, scaled image. This is convenient when system bandwidth is limited and you don't require any precision measurements on the image.

The other use of the term "resolution" is commonly found in data acquisition applications and refers to the number of quantization levels used in A/D conversions. Higher resolution in this sense means that you would have improved capability of analyzing low-contrast images. This resolution is specified by the A/D converter; the frame grabber determines the resolution for analog signals, whereas the camera determines it for digital signals (the frame grabber must have the capability of supporting whatever resolution the camera provides, though).

Camera Type

Specialized applications may require different types of sensors. Although we normally think of a camera as something taking a photograph, any signal can be displayed as an image, as long as it can be put into a format that is compatible with the frame grabber. The typical camera uses a charge-coupled device (CCD) array, but any array of detectors that produce voltage outputs can be treated as a camera, as long as the proper synchronization signals are associated with it. A single detector moved in a scanning pattern over an area can also be used to produce a video signal. Other detectors commonly used in imaging applications include thermal or infrared cameras and scanning microscopes.

Additional Features

There are many other camera features available. Many cameras support programmatic control of various parameters via a serial interface. Some examples include:

  • Shutter control
  • Exposure time settings
  • Different acquisition modes
  • Triggering features
  • Focus control
  • Strobe outputs

National Instruments digital frame grabbers have an integrated serial interface that frees up the computer's serial port and allows for extremely simple implementation of programmatic camera control functionality.
Related Links:
Camera Advisor (camera selection guide)