Author Affiliations
1College of Information Engineering, Pingdingshan University, Pingdingshan, Henan 467000, China2Institute of Surveying and Mapping, Information Engineering University, Zhengzhou, Henan 450001, Chinashow less
Fig. 1. Results comparison. (a) Operation results by PST; (b) Operation results by Eq. (13)
Fig. 2. Error curve of the calculation result in Fig.1
Fig. 3. Grayscale variance and angle map variance maps of the source images. (a)(d) Blurred and clear source images;(b)(e) grayscale variance of Fig. (a) and (d); (c)(f) angle map variance of Fig. (a) and (d)
Fig. 4. Original image and high-frequency information extraction results. (a)(b) Original images; (c)(d) high-frequency information extracted by kernel function in Ref. [9] and proposed kernel function
Fig. 5. Flow chart of proposed multi-focus image fusion algorithm
Fig. 6. Fusion results of Lytro-07. (a) Fusion result by DWT; (b) fusion result by Laplace operator; (c) fusion result by SR; (d) fusion result by guided filter; (e) fusion result by JCAN; (f) fusion result using the kernel function in Ref. [9]; (g) fusion result using the proposed kernel function
Fig. 7. Difference maps of fusion results. (a)-(g) Corresponding to the difference maps in Fig. 6 (a)--(g)
Fig. 8. Local enlarged drawing. (a)(b) Corresponding to the local enlarged drawing in Fig. 7 (f) and (g)
Fig. 9. Source images and fusion results of proposed algorithm. (a)(b) Source images; (c) fusion results
Source image | Variance ofblurred images | Variance ofclear images | Variance of anglemap of blurred images | Variance of anglemap of clear images |
---|
1 | 409.11 | 584.31 | 972.72 | 1031.26 | 2 | 907.29 | 967.25 | 1210.74 | 1318.45 |
|
Table 1. Grayscale variance and angle map variance of the blurred and clear images
Sourceimage | Method | Mutualinformation | Averagegradient | Informationentropy | SpatialFrequency | Structuralsimilarity | Correlationcoefficient |
---|
Lytro-07 | DWT | 1.2596 | 2.081 | 7.725 | 7.9721 | 0.8505 | 0.9846 | Laplace | 1.2814 | 2.085 | 7.7265 | 7.8191 | 0.8518 | 0.9844 | SR | 1.3125 | 2.097 | 7.7299 | 8.0113 | 0.8558 | 0.9847 | Guided filter | 1.3261 | 2.090 | 7.7279 | 7.9890 | 0.8627 | 0.9842 | JCAN | 1.3274 | 2.112 | 7.7322 | 8.1343 | 0.8841 | 0.9871 | Kernel function of Ref. [9] | 1.3275 | 2.110 | 7.7356 | 8.0891 | 0.8808 | 0.9849 | Proposed algorithm | 1.3268 | 2.116 | 7.7357 | 8.0958 | 0.8958 | 0.9852 | Lytro-01 | DWT | 1.222 | 2.320 | 7.866 | 6.8799 | 0.8419 | 0.9558 | Laplace | 1.221 | 2.440 | 7.995 | 6.7878 | 0.8421 | 0.957 | SR | 1.4103 | 2.529 | 7.9431 | 6.9071 | 0.8711 | 0.9515 | Guided filter | 1.4151 | 2.502 | 7.923 | 6.9079 | 0.8902 | 0.9561 | JCAN | 1.4260 | 2.660 | 8.0842 | 7.1654 | 0.8992 | 0.9625 | Kernel function of Ref. [9] | 1.251 | 2.610 | 7.985 | 7.1644 | 0.8524 | 0.9578 | Proposed algorithm | 1.422 | 2.630 | 8.1125 | 7.1676 | 0.8731 | 0.959 | Lytro-03 | DWT | 1.215 | 1.950 | 7.6214 | 6.5813 | 0.879 | 0.9864 | Laplace | 1.216 | 1.980 | 7.5931 | 6.517 | 0.8802 | 0.9862 | SR | 1.2129 | 2.013 | 7.6197 | 6.5045 | 0.8812 | 0.9869 | Guided filter | 1.2224 | 2.010 | 7.6258 | 6.6014 | 0.8865 | 0.9865 | JCAN | 1.2487 | 2.070 | 7.6272 | 6.5554 | 0.8873 | 0.9875 | Kernel function of Ref. [9] | 1.219 | 2.030 | 7.6317 | 6.5239 | 0.8817 | 0.9882 | Proposed algorithm | 1.223 | 2.040 | 7.6328 | 6.5593 | 0.8953 | 0.9885 | Lytro-12 | DWT | 1.436 | 3.241 | 7.4608 | 11.6537 | 0.8715 | 0.9865 | Laplace | 1.441 | 3.247 | 7.4335 | 11.5088 | 0.8801 | 0.9869 | SR | 1.4329 | 3.265 | 7.4667 | 11.6611 | 0.9114 | 0.9876 | Guided filter | 1.4343 | 3.270 | 7.4614 | 11.6723 | 0.9025 | 0.9873 | JCAN | 1.468 | 3.280 | 7.4733 | 11.6942 | 0.9170 | 0.9889 | Kernel function of Ref. [9] | 1.443 | 3.283 | 7.4721 | 11.6762 | 0.8767 | 0.9893 | Proposed algorithm | 1.446 | 3.286 | 7.4726 | 11.6973 | 0.9135 | 0.9892 | Lytro-14 | DWT | 1.337 | 2.810 | 7.5752 | 9.8157 | 0.8871 | 0.9692 | Laplace | 1.339 | 2.820 | 7.5598 | 9.8079 | 0.8874 | 0.9698 | SR | 1.3423 | 2.830 | 7.5601 | 9.8547 | 0.8883 | 0.9694 | Guided Filter | 1.341 | 2.830 | 7.5576 | 9.8528 | 0.889 | 0.9692 | JCAN | 1.347 | 2.840 | 7.5704 | 9.8552 | 0.8897 | 0.9771 | Kernel function of Ref. [9] | 1.344 | 2.860 | 7.5609 | 9.8047 | 0.8904 | 0.9700 | Proposed algorithm | 1.343 | 2.870 | 7.5606 | 9.8717 | 0.8910 | 0.9703 |
|
Table 2. Objective evaluation indexes of fusion results