(2) Discussion on smoothing method 
(J) Smoothing matrix 
As a result of comparison of2x2-7x7 matrices, it was shown that 3x3 matrix is most 
appropriate, because the adequate smoothing effect is obtained. The histogram pattern after 
smoothing showed that it is effective method to reduce the discontinuity of image data, because 
the spike shaped noises had been removed adequately. Figure 4 shows the comparison of 
histograms before and after applying 3x3 matrix. 
(a) Original data (b) after smoothing 
Fig. 4: Comparison of histograms before and after smoothing(3x3) 
© ABIC minimization method 
It is clear from the statistical values in Table 3 that the values of original temperature 
distribution data are maintained and smoothing effect is obtained adequately. The calculation 
results was obtained by applying the ABIC minimization method to temperature distribution 
data. Figure 5 shows the comparison of histogram before and after applying ABIC minimization 
method(1991 July 21). However, it has remained the problem that huge time is necessary for 
the calculation (about 6 hours for 64 x 64 pixels). 
Table 3:The statistical values of the original and after calculation by ABIC 
1991 Ju 
ly 21 
1991 Sep. 2 
1992 Sep. 1 
Original 
ABIC 
Original 
ABIC 
Original 
ABIC 
Maximum 
209 
207 
211 
189 
188 
186 
Minimum 
114 
129 
160 
150 
139 
143 
Mean 
187.796 
187.733 
177.971 
177.811 
164.657 
164.456 
Std.Dev. 
13.769 
12.453 
6.171 
5.07 
9.384 
8.535