Evaluation
HAF and the algorithm proposed in MATLAB R2005a
were implemented and tested and the results of their
application on Lena image are provided here as example.
Figure 3 shows the main noisy image and Figure 4
presents the outputs after applying the three HAF, MF,
and proposed algorithms. Analysis of the results show the
medium filters of the fuzzy adaptor bring much less
opaqueness to the image in comparison to normal
medium filters, and this holds true for the offered filter as
well. It could be said with certainty about HAF algorithm
that there are conditions other than noise elimination that
apply noise to the image. This has been witnessed in
different tests on various images (Figure 4b). A precise
theoretical HAF analysis will lead us to this point as well.
The proposed algorithm produces better results in than
HAF and despite the fact that RB and Mfs are the same
for both of them, the proposed algorithm does not
produce any noise in the image which is due to the
powerful adaptor we have presented for it serving to find
the best possible state for all Mf parameters through an
evaluation function, and shows the final output based on
it, unlike HAF which gave values to said parameters
based on the image histogram. The only problem of the
proposed method is its need for more time as compared to
HAF. It can be said that the proposed method is at least
256 times slower than HAF, and this is obvious for HAF
estimates the Mf parameters and applies fuzzy filters only
once, while the proposed method applies the filters for all
possible states of Mf parameters and recognizes the best
output through an evaluation function in which case
minimum number of possible cases is 256 (assume that at
least one parameter may change among gray levels of 0
to 256, while taking into account all parameters we will
see that the number of states are far more than this).
Evaluation
HAF and the algorithm proposed in MATLAB R2005a
were implemented and tested and the results of their
application on Lena image are provided here as example.
Figure 3 shows the main noisy image and Figure 4
presents the outputs after applying the three HAF, MF,
and proposed algorithms. Analysis of the results show the
medium filters of the fuzzy adaptor bring much less
opaqueness to the image in comparison to normal
medium filters, and this holds true for the offered filter as
well. It could be said with certainty about HAF algorithm
that there are conditions other than noise elimination that
apply noise to the image. This has been witnessed in
different tests on various images (Figure 4b). A precise
theoretical HAF analysis will lead us to this point as well.
The proposed algorithm produces better results in than
HAF and despite the fact that RB and Mfs are the same
for both of them, the proposed algorithm does not
produce any noise in the image which is due to the
powerful adaptor we have presented for it serving to find
the best possible state for all Mf parameters through an
evaluation function, and shows the final output based on
it, unlike HAF which gave values to said parameters
based on the image histogram. The only problem of the
proposed method is its need for more time as compared to
HAF. It can be said that the proposed method is at least
256 times slower than HAF, and this is obvious for HAF
estimates the Mf parameters and applies fuzzy filters only
once, while the proposed method applies the filters for all
possible states of Mf parameters and recognizes the best
output through an evaluation function in which case
minimum number of possible cases is 256 (assume that at
least one parameter may change among gray levels of 0
to 256, while taking into account all parameters we will
see that the number of states are far more than this).
การแปล กรุณารอสักครู่..
Evaluation
HAF and the algorithm proposed in MATLAB R2005a
were implemented and tested and the results of their
application on Lena image are provided here as example.
Figure 3 shows the main noisy image and Figure 4
presents the outputs after applying the three HAF, MF,
and proposed algorithms. Analysis of the results show the
medium filters of the fuzzy adaptor bring much less
opaqueness to the image in comparison to normal
medium filters, and this holds true for the offered filter as
well. It could be said with certainty about HAF algorithm
that there are conditions other than noise elimination that
apply noise to the image. This has been witnessed in
different tests on various images (Figure 4b). A precise
theoretical HAF analysis will lead us to this point as well.
The proposed algorithm produces better results in than
HAF and despite the fact that RB and Mfs are the same
for both of them, the proposed algorithm does not
produce any noise in the image which is due to the
powerful adaptor we have presented for it serving to find
the best possible state for all Mf parameters through an
evaluation function, and shows the final output based on
it, unlike HAF which gave values to said parameters
based on the image histogram. The only problem of the
proposed method is its need for more time as compared to
HAF. It can be said that the proposed method is at least
256 times slower than HAF, and this is obvious for HAF
estimates the Mf parameters and applies fuzzy filters only
once, while the proposed method applies the filters for all
possible states of Mf parameters and recognizes the best
output through an evaluation function in which case
minimum number of possible cases is 256 (assume that at
least one parameter may change among gray levels of 0
to 256, while taking into account all parameters we will
see that the number of states are far more than this).
การแปล กรุณารอสักครู่..