Biomedical Engineering Reference
In-Depth Information
Color processing system
Sensor
Processor
Photosensitive layer
Preprocessing layer
Linear/nonlinear
Color coordinate transformation
Input
Output
Simple feature extraction
Redundancy reduction
Spectral image
Entropy minimization
Computation of sensor parameters
FIGURE 16.3
Model of the color-processing system.
16.4.2
Preprocessing Layer
We are now interested in color processing. The preprocessing layer is designed to learn
color space for the responses of photosensitive layer. The learning is done by training self-
organizing feature map (SOM), which yields a grouping of similar colors.
The preprocessing layer may operate in two modes: off- and online mode, respectively.
The purpose of the offline or learning mode is to adjust the internal parameters of the pre-
processing layer. Training the SOM is the offline mode of operation of the preprocessing
layer. In the online mode, the sensor is presented with arbitrary colors and it responds
with their color coordinates in learned color space.
16.5
Imaging with Bacteriorhodopsin
To show the potential of BR in photosensing, we have prepared single-pixel photosensors
based on wild-type BR and its analogs (27,28). We selected polyvinylalcohol (PVA) as the
polymer matrix to prepare the thick films, which is supported by the following facts:
The preparation process is straightforward and relatively simple (29).
The PVA matrix forms a stable environment for the purple membrane (PM) frag-
ments containing BR (30).
The PVA matrix permits measurements concerning the alternating-current pho-
toelectric response (PER) and optical properties of the film (6).
The films are mechanically stable.
The preparation method has a disadvantage. If direct control of the homogeneity of the
films or the orientation of the PM fragments in the films is desired, the preparation process
becomes much more difficult.
Search WWH ::




Custom Search