Image processing Essay

Where Xi
t is the pollinator solution vector Xi at iteration t and g is considered the current best solution. The parameter L is the strength of the pollination which is considered
as the step size randomly based on levy distribution.
Then, based on rule 2 and 3, local pollination can be expressed as
Xt+1
i = Xit + (Xtj Xk t) (2)
Where Xt
j and Xk t are solution vectors of a different flower of the same plant. The parameter  is varying in the range from 0 to 1 established on the uniform distribution.
Most the application can be switching between the global and local pollination based
on the rule 4 and p= 0.8 is more efficient.
2.2 The Stationary Wavelet Transform
The discrete wavelet transform is widely used in the image processing. It provides
a framework for the image decomposition. In each level the image is decomposed
into subbands a coarser resolution or lower frequency band. The DWT is not a timeinvariant transform. Without shift-invariance, slight shifts in the input image will generate variations in the wavelet coefficients that should cause the artifacts in the reconstructed image [17]. Shift-variance is proceeded by the decimation process, and can be
resolved by using the un-decimation algorithm. The way to overcome this disadvantage. There is another version of DWT, called decimated DWT. The stationary Wavelet
Transform (SWT) is comparable the DWT but the only process of downsampling is
overcome that indicate SWT is a Translation invariant.The SWT is based mainly on the
concept of no decimation. the SWT is a perfect shift-invariant transform. That is done
by suppressing the down-sampling step of the decimated algorithm and instead of upsampling the filters by inserting zeros between the filter coefficients. In the decimated
algorithm, the filters are used for the rows at the first and then for the columns [18]. In
this case, however, although the two images (Pan, MS) produced (one approximation
and three detail images) are at half the resolution of the original. They have the same
size of the original image. The approximation images from the undecimated algorithm
are therefore indicated as parallelepiped levels, with the spatial resolution becoming
coarser at each higher level and the size still the same. The undecimated algorithm is
redundant, meaning some detail information may be kept in adjacent levels of transformation. It also needs more space to store the results of each level of transformation and,
in spite of it is shift-invariant. It does not resolve the problem of feature orientation. A
previous level of approximation, resolution J1, can be reconstructed exactly by applying the inverse transform to all two images at resolution J2 and combining the resulting
images. Basically, the inverse transform involves the same steps as the forward transform, but they are implemented in the reverse order. In the decimated case, this means
up-sampling the approximation and detail images and using the reconstruction filters,
which are inverses of the decomposition scaling and wavelet filters, first by columns
and then by rows.