Page 113 [113]
4) Noise reduction, antialiasing, and sharpening. Problems can arise with very small details
in an image. If the detail is only captured on a red-sensing pixel or a blue-sensing pixel,
the RAW converter may find it difficult to assign a colour to that pixel. Simple demosaicing
methods are also not proficient at maintaining edge detail, so most RAW converters also
perform some combination of edge-detection and antialiasing to avoid colour artefacts,
noise reduction, and sharpening.
In general, RAW converters perform all of these tasks, but they may use very different
algorithms to do so, which is why the same image may look guite different when processed
through different RAW converters. If the captured image is converted to a J PEG file by the
camera, a RAW converter built into the camera carries out all the tasks listed above to turn
the RAW capture into a colour image, and then compresses it using JPEG compression
(Figure 3-1(a)). Some cameras allow parameters to be set for this conversion, typically a
choice of SRGB or Adobe RGB as colour space, a sharpness value, and perhaps a tone
curve or contrast setting but it is difficult to adjust these parameters on an image-by-image
basis, and generally the JPEG produced is locked into the camera’s interpretation of the
scene.
Image capture in RAW format and conversion to a] PEG or TIFF file from the RAW file by
external RAW conversion software (Figure 3-1(b)) is therefore recommended, as this returns
control over the interpretation of the image to the user by allowing all the aforementioned
aspects of the conversion to be adjusted. The only on-camera settings that have an effect on
the captured pixels are the ISO speed, the shutter speed, and the aperture setting. This
allows all the settings such as contrast, saturation, tone curves typically employed by the
camera for aesthetic enhancement and which distort the sensor data, to be turned off.
In addition, conversion by external RAW conversion software also allows control over the
tonal information contained per pixel. In RAW format almost all cameras capture at least 12
bits, or 4096 shades, of tonal information per pixel. The J PEG format, however, is limited to
8 bits per channel per pixel. This means that the camera’s built-in RAW converter throws
away a large amount of the captured data with essentially no control by the user over what
gets discarded. In order to access all 12 bits of the original RAW file, the data should
preferentially be converted to a 16-bit TIFF file.
Finally, on conversion the external RAW conversion software automatically produces a
unique ICC profile which records all the metadata (including the gamma correction applied to
the image) describing this conversion (Figure 3-2). This profile is contained within the TIFF
file and can be used by the post-processing software to recover the linear data observed by
the camera sensor. The transformation from the converted TIFF file to linear data using the
ICC profile has been integrated into the various image correction workflows in the nip2
workspace.
Version No. 1.0 105 Date : 14/10/2013