Graphics Reference
In-Depth Information
comparable, unified rules are adopted in terms of integration parameters selection. In
the decomposition of pyramid, high-frequency coefficients take the maximum and the
low-frequency coefficients take the average. When the images are in transformation,
layers are all taken as 4. In the spatial frequency fusion algorithm, different types of
test images lead to different sizes of the best Block. According to subjective effects of
fused images by experiments, in this essay, we select the best Block size of multiple
images to be 8
, and the threshold
TH to be 1. The dictionary of multi-focus image
fusion algorithms based on sparse representation and orthogonal matching pursuit has
used over-complete DCT dictionary and K-SVD dictionary, and the size of the image
block is 8
×
8
×
8
.
3.2
Image Fusion Results and Subjective and Objective Evaluations
(1) Subjective and objective evaluations of each image fusion algorithm to the multi-
focus image of group A
Original multi-focus image data and fusion results of group A are shown in figure
2, and objective evaluation indicator data are shown in table 1.
(a) (b) (c) (d) (e)
(f) (g) (h) (i)
Fig. 2.
Experimental Results of Multi-focus Image Fusion of Group A. (a)Foreground Fuzzy
Image (b)Background Fuzzy Image (c)8*8 DCT (d)8*8 K-SVD (e)FSD Algorithm (f)Contrast
Algorithm (g)DWT Algorithm (h)SIDWT Algorithm (i)Spatial Frequency Algorithm.
Table 1.
Objective Evaluation Indicator Data to Multi-focus Fusion Image of Group A
Q
AB/F
Algorithm
MI
AG_A
AG_B
AG_F
CC_A
CC_B
DD_A
DD_B
8*8DCT
7.903 2.550
1.827 2.790
0.991
0.983
0.946
2.332
0.698
8*8KSVD
7.467 2.550
1.827 2.662 0.991 0.983 1.007
2.125
0.683
FSD
6.231
2.550
1.827
2.830
0.989
0.989
2.124
2.172
0.663
Contrast
6.873
2.550
1.827
3.747
0.993
0.982
2.159
2.199
0.630
DWT
6.172
2.550
1.827
3.847
0.989
0.979
2.629
2.243
0.604
SIDWT
6.627
2.550
1.827
3.661
0.992
0.982
2.675
2.262
0.671
SF
8.058
2.550
1.827
3.452
0.993
0.983
2.225
2.212
0.685
Search WWH ::
Custom Search