îiÊft fT 
I SMODI 
JjäH§ J «U i 
«tassa ! taxas I 
3.2.2 The calculation principle and method of texture 
feature value 
The principle on extraction of texture feature image is that sub 
image’s formed by each small window is firstly used to obtain 
texture features value by calculating sub-image gray level co 
occurrence matrix and the texture feature value denoting this 
window is evaluated to the window central point. This has 
completed the texture features computation of the first sub 
image. Then the window is moved a pixel and forms another 
small window image. Next, the author compute texture features 
value of new sub-image repeatedly which is evaluated to the 
central point of this window. In turn, whole image will form a 
texture feature value matrix composed by the texture feature 
value which is transformed to the texture feature image. 
Figure 4 is the principal of texture feature calculation with the 
5x5 sliding window as an example. 
WntfJWiflW 
i. a tn№isr№» 
Figure 4. The principle of texture feature calculation 
(based on 5x5window) 
3.2.3 The procedure and result of texture feature images 
of the original image 
According to the theory and study, we developed the program 
of texture feature extraction by means of VC++. The function 
of this procedure is to have a texture analysis and output texture 
analysis for input sensing image. The interface of extract 
texture feature image is figure 5. 
Figure 6. Texture feature image of contrast 
Figure 7. Texture feature image of correlation 
Figure 8. Texture feature image of entropy 
Figure 5. The interface of extract texture feature image 
The function of this procedure is clear with simple operation 
and strong specialization. We obtain the texture characteristic 
images through the procedure. 
The texture feature images are obtained through 5 X 5 and 7 X 7 Figure 9. Texture feature image of second moment 
sliding windows computing the texture feature value all over 
the images. There are as follows: 
llll 
llll 
llll 
II Ilf 
IUIII 
Hill