Technical details of line scan cameras
The other standard variable is the pixel size which is often overlooked when selecting line scan, but can have a big effect in applications that are light limited. In addition specialist line scan cameras with dual-line or TDI are available, which will be discussed later in this section.
The sensor resolution is a measure of the number of pixel elements that the sensor has. The more pixels of a given pixel size, the longer the physical length of the sensor and hence a larger format lens will be required. Most sensors up to 1K pixels use standard C-mount lenses.
When the sensor is 2K pixels or more, large format lenses are normally required, such as the F-mount type, which has a sufficiently large image circle to image onto the sensor without significant vignetting. As mentioned in the lens section of this handbook, vignetting can be a serious problem in line scan applications, as it causes a non-linear response across the sensor.
The pixel clock rate is the speed at which pixel values are read out of a camera. To reach a high line frequency on a higher resolution camera (i.e. 8K pixels at 80 kHz), image data is transmitted at high frequency in parallel over multiple channels. For this reason line scan cameras always use a high-speed digital output, as analogue bandwidth is too limited. Typically, the pixel clock provides a fixed rate when transferring the pixel charges. While pixels are transferred with a fixed frequency, the line frequency can be variable or constant. Running the camera in 'freerun' means that lines of video are continuously output at the same rate. A constant line rate is no problem as long as the object being imaged is moving at constant speed.
In many applications however, the speed of the moving object may vary requiring the use of what is called a 'line drive' or EXSYNC signal. This is a trigger signal that is generated at regular spatial intervals (referenced to the object), so that the scanned lines are synchronous with movement of the objects. This is usually achieved with an encoder. In setting up such a system care has to be taken that the maximum rate of the line-drive signal (encoder output) is limited by the speed of the camera and the number of pixels (length of the sensor)
As the images above show, an image of a circle captured without an encoder to synchronise the movement with the line speed results in a distorted image, as the object speed varies during the scan. Square, uniformly geometric pixels in the resulting image are desirable as this simplifies interpretation, analysis and display of the data. In practice, this is the way most systems are designed.
A practical example to calculate line frequency and exposure time
- Object width: B = 370 mm
- Object speed: v = 3 m/s
- Required resolution: Dx = 0.2 mm/pixel
The number of pixels required is calculated as follows:
n Pixel = B / Dx
Thus a resolution of minimum 1850 pixels is required. By default models are normally available with 2048 pixels. If an object width of 370 mm is imaged on the 2k-line, one pixel captures 0.18 mm of the object. Assuming the horizontal and vertical resolution should be even (ratio 1:1), the line frequency can be calculated as follows:
Fz = v / Dy
Due to the calculated resolution Dx = Dy with Dx = 0.18 mm the result adds up to:
Fz = 16,667 Hz
The line scan camera should be able to operate at a line frequency of minimum 16.7 kHz.