Mitsubishi P91DW User Manual Page 91

  • Download
  • Add to my manuals
  • Print
  • Page
    / 206
  • Table of contents
  • BOOKMARKS
  • Rated. / 5. Based on customer reviews
Page view 90
Alpha Innotech User’s Manual, version 4
4-16
General Information on How Filters Work
To enhance an image, filters change the value assigned to each pixel. The new value
assigned to a pixel is determined based on the values of the other pixels in its local
vicinity (or “neighborhood”). The neighborhood is a two-dimensional matrix of pixel
values, where each dimension has an odd number of elements. The "pixel of interest" is
the one at the center of the neighborhood. This is the pixel whose old value is being
replaced with a new one as the result of the filtering algorithm.
The pixels in a neighborhood provide information about the brightness trend. This
information is important to the filtering process. The brightness trend is also referred to
as the "spatial frequency." Images with high spatial frequency content contain large,
closely spaced changes in pixel values. For example, on a black and white checkerboard,
the smaller the squares, the higher the frequency content.
Images with low spatial frequency content (for example, images of clouds) contain large
areas of slowly changing pixel values.
Most of the filter options available (with the exception of the Noise filters) use a
weighted summation process to determine the value assigned to the pixel of interest. Each
pixel in a 3x3 neighborhood is multiplied by a "convolution kernel" having the same
dimensions. The resulting sum is assigned to the pixel of interest.
(K1xP1)+
(K2xP2)+
(K3xP3)+
(K4xP4)+
P1 P2 P3 K1 K2 K3 (K5xP5)+
P4 P5 P6 X K4 K5 K6 (K6xP6)+
P7 P8 P9 K7 K8 K9 (K7xP7)+
(K8xP8)+
3x3 pixel neighborhoodconvolution kernel (K9xP9)
(P5 is being calculated) New Value for P5
Each element of the convolution kernel is a weighting factor, also called a "convolution
coefficient." The size and arrangement of these weighting factors determine the type of
transformation the image will undergo. Changing a weighting factor influences the
overall sum and, therefore, affects the value given to the pixel of interest.
Page view 90
1 2 ... 86 87 88 89 90 91 92 93 94 95 96 ... 205 206

Comments to this Manuals

No comments