WÊKÊÊKÊÊHMH 
thod utilizes 16 neighboring values of the input intensity 
I(u+m,v+n) to determine the input intensity I(x,y) 
I(x,y) = £ a I(u+m,v+n), -1< m,n <2, 
J m,n m,n 
where (m,n) is the pixel location of the input signal and 
a corresponds to the resampling filter in two-dimensions. 
TÏÏé n one-dimensional resampling filter a m produces the cubic 
convolution polynomial 
1 - 2m^ +ImI 3 
0$ImI<1; 
4 - 8 I m I +5m^ -Iml^, l^|m|<.2; 
(8) 
0, 
2$ Imi 
Histogram of the grey levels. In order to evaluate the re 
sampling effect on the output image, the histogram of density 
number in grey levels are shown in Figurel-1 throughl-4 in band 
6. It is seen that the histogram of the uncorrected data re 
sembles well to that of the resampled data via NN, whereas 
the histogram of resampled data via BI and CC are somewhat 
dissimilar to that of the uncorrected data. 
Fig. 1 Histogram of original and resampled data