cfa_to_rgb — Convert a single-channel color filter array image into an RGB image.
cfa_to_rgb converts a single-channel color filter array
CFAImage into an RGB image
filter array images are typically generated by single-chip CCD
cameras. The conversion from color filter array image to RGB image
is typically done on the camera itself or is performed by the device
driver of the frame grabber that is used to grab the image. In some
cases, however, the device driver simply passes the color filter
array image through unchanged. In this case, the corresponding
HALCON frame grabber interface typically converts the image into an
RGB image. Hence, the operator
cfa_to_rgb is normally used
if the images are not being grabbed using the HALCON frame grabber
grab_image_async), but are
grabbed using function calls from the frame grabber SDK, and are
passed to HALCON using
In single-chip CCD cameras, a color filter array in front of the
sensor provides (subsampled) color information. The most frequently
used filter is the so-called Bayer filter. The color filter array
has the following layout in this case:
Each gray value of the input image
CFAImage corresponds to
the brightness of the pixel behind the corresponding color filter.
Hence, in the above layout, the pixel (0,0) corresponds to a green
color value, while the pixel (0,1) corresponds to a blue color
value. The layout of the Bayer filter is completely determined by
the first two elements of the first row of the image, and can be
chosen with the parameter
CFAType. In particular, this
enables the correct conversion of color filter array images that
have been cropped out of a larger image (e.g., using
The algorithm that is used to interpolate the RGB values is
determined by the parameter
Interpolation = 'bilinear', a bilinear
interpolation is performed. While this algorithm is very fast, it
typically leads to “zipper-like” artifacts and color artifacts at
strong edges. For
'bilinear_dir', a modified version of the bilinear
interpolation is performed that may lead to fewer zipper-like
artifacts, especially at horizontal or vertical edges in the image.
The results may still exhibit color artifacts at strong edges,
however. The runtime of this algorithm is only slightly longer than
that of bilinear interpolation. For
'bilinear_enhanced', an enhanced version of the bilinear
interpolation is performed that produces fewer zipper-like artifacts
and color artifacts than the other two bilinear algorithms in most
cases. The runtime of this algorithm is significantly longer than
that of the other two algorithms.
If 'mmx_enable' is set to 'true' (and the SIMD
instruction set is available), the internal calculations for byte
images are performed using SIMD technology for
Interpolation = 'bilinear' and
Interpolation = 'bilinear_dir'.
Interpolation = 'bilinear' and
Interpolation = 'bilinear_dir',
can be executed on OpenCL devices. The width of the input image should be
a multiple of four for byte images, or two for uint2 images, as the
operation will be much slower otherwise.
→object (byte* / uint2*) *allowed for compute devices
→object (byte / uint2)
Color filter array type.
Default value: 'bayer_gb'
List of values: 'bayer_bg', 'bayer_gb', 'bayer_gr', 'bayer_rg'
Default value: 'bilinear'
List of values: 'bilinear', 'bilinear_dir', 'bilinear_enhanced'
cfa_to_rgb returns 2 (H_MSG_TRUE) if all parameters are correct. If
the input is empty the behavior can be set via
set_system(::'no_object_result', <Result>:). If necessary,
an exception is raised.