PL-A780 Camera Specifications
7
PixeLINK PL-A780
Document No.: 04646-01
MACHINE VISION CAMERA
SYSTEM GUIDE
Copyright © 2004 PixeLINK
All Rights Reserved
3.2 Sensor
The sensor used in the PL-A780 cameras is a Fill Factory IBIS4-6600. The IBIS4-6600 is a
6.6 megapixel solid-state CMOS image sensor with 2210 x 3002 active pixels. The image
size is fully programmable to a user-defined ROI. Pixels are on a 3.5 µm pitch.
The color version of the sensor uses a Bayer pattern to generate the RGB information. This
is a common approach to detect color with a single sensor and is used extensively in the
industry. The Bayer pattern consists of color filters placed over each pixel in alternating
green/red on the odd rows and blue/green on the even rows. To determine the RGB color of
any particular pixel, a Bayer to RGB conversion is performed where the color information of
neighboring pixels is used to determine the missing components of the RGB color for a
particular pixel. The Bayer to RGB conversion has the effect of a smearing filter and slightly
reduces the effective resolution or resolving power of the sensor. It also can introduce
colored artifacts into an image on the edges of high contrast areas. Resampling the image,
as discussed below, can reduce both of these effects.
The monochrome version of the sensor behaves in the same manner as the color version for
ROI, pixel addressing and resampling. The monochrome version does not use the Bayer
pattern and Bayer to RGB conversions are not applied.
3.2.1 ROI, Pixel Addressing and Resampling
The implementation of the sensor in the PL-A780 limits the active area to 2208 x 3000.
Within this Full Field of VIEW (FFOV), the ROI size and position has a granularity of 24
pixels. The minimum size of the horizontal or vertical dimensions of the ROI is 24. Possible
ROI sizes are 24 x 24, 24 x 48, 48 x 48, 48 x 72, … 2208 x 3000. The top left corner of the
ROI can be positioned on any row and column that is a multiple of 24.
The resolution can be reduced while maintaining the field of view by subsampling, (referred to
as decimation), resampling, binning or averaging blocks of pixels. Decimation is performed
by skipping blocks of pixels. The smallest block is a 2 x 2 square or four pixels. No
decimation has a factor of 1 meaning every pixel block is used. To reduce the resolution by a
factor of 2, every other block in a row of blocks is used and every other row of blocks is
skipped. The final resolution will be ¼ of the pixels in the ROI. Decimation of 3 will skip two
blocks of pixels.
The ROI can be decimated by factors of 1, 2, 3, 4 and 6. This will reduce the image size by
factors of 1, 4, 9, 16 and 36 respectively. For example, with a decimation mode of 6, the
FFOV ROI (2208 x 3000, 6.6 megapixels), can be reduced to 368 x 500 or 184,000 pixels
while maintaining the FFOV. As the resolution and ROI are reduced, the frame rate
increases as shown in Table 5 above (page on page 6).
The following figures demonstrate the effects of decimation and how the 2 x 2 pixel blocks
are determined.
