Image Processing Reference
In-Depth Information
Signal processing
Storage
AFE
CDS
Canceling eciency
AGC
Amp. noise
ADC
Quantization noise
Digital signal
Raw data
Defect correction
DSP
Demosaicking
White balance
Color correction
Noise reduction
TIFF, JPEG,
etc.
/cancellation
Gamma correction
Other corrections
Light makeup
Raw
data
Image
data
Aberration
(chromatic)
(distortion)
Shading
Heavy, but
modest makeup
RAW converter
Compression
FIGURE 8.4
Image information and elements of quality: signal processing.
it is almost impossible to discuss camera performance by referring to only sensor perfor-
mance without referring to functions and performances of DSPs. Even in the case of raw
data output, it is quite rare for raw data to be output without any basic correction such
as defect correction. Data that are finely converted to formats such as TIFF and JPEG are
stored in memory.
As observed above, there are many elements that influence information quality. It is
important to comprehend the degree, influence, and priority of these elements.
8.2 Signal Processing
To return to a subject mentioned in Section 8.1, what DSPs should originally process is
described. Digitization of the sensor output signal alone cannot make it a color image
signal the human eye can process, but various processing operations are necessary. An
example outline of the overall process follows.
Each color signal, R, G, and B, from the sensor output is selected and used to build an
RGB color image. The signal is then corrected to easily viewable brightness, adjusted to a
natural color with enhancement of resolution and contrast, and converted to a format such
as TIFF or JPEG.
An example of processing flow is shown in Figure 8.5. In the first stage, linear opera-
tion of pixel data is carried out; after corrections of defect, brightness, and white balance,
demosaicking and color conversion follow. Nonlinear processing follows, such as color/
tone conversion, noise reduction, and edge enhancement. A series of these types of signal
processing is an important operation that has a big impact on the final image quality or
impression. This is called the “image processing engine” and camera manufacturers put
great effort into it.
