Author Affiliations
1Faculty of Information Engineering and Automation, Kunming University of Science and Technology, Kunming 650500, Yunnan , China2Yunnan Key Laboratory of Artificial Intelligence, Kunming University of Science and Technology, Kunming 650500, Yunnan , Chinashow less
Fig. 1. Flow chart of proposed algorithm
Fig. 2. Color correction results.(a) original images; (b) pseudo color images of original images;(c)color correction images;(d)pseudo color images based on color correction
Fig. 3. Comparison of different wavelet decomposition levels. (a) Original image; (b) level 0; (c) level 2; (d) level 4; (e) details of level 4
Fig. 4. Comparison of underwater optical image clarity effect. (a) Original image; (b) dark channel prior dehaze; (c) multiscale wavelet dehaze
Fig. 5. 2D gamma correction histogram contrast map. (a) Original image; (b) illumination component map of original image; (c) histogram before correction; (d) 2D gamma correction map; (e) illumination component map of 2D gamma correction image; (f) histogram after correction;
Fig. 6. Comparison of sharpen. (a) Original image; (b) gray scale image of original image; (c) edge detail of original image; (d) sharpened image; (e) gray scale image of sharpened image; (f) edge detail of sharpened image
Fig. 7. 2D gamma weightmap. (a) 2D gamma correction result;(b)luminance weightmap;(c)chromatic weightmap;(d)saliency weightmap
Fig. 8. Sharpen weightmap. (a) Sharpen result; (b) luminance weightmap;(c)chromatic weightmap;(d)saliency weightmap
Fig. 9. Underwater optical image enhancement results based on different algorithms. (a) Original image;(b)algorithm in reference[
6];(c)algorithm in reference[
7];(d)algorithm in reference[
8];(e)algorithm in reference[
9];(f)proposed algorithm
Fig. 10. Application test results. (a) Original image matching results; (b) enhanced image matching results
Fig. 11. Evaluation results of low illumination image
Fig. 12. Canny edge detection comparison results. (a) Original image; (b) Canny operator detection results of original image; (c) enhanced image; (d) Canny operator detection results of enhanced image
Image | Algorithm in reference[9] | | Proposed Method |
---|
UCIQE | UIQM | AG | UCIQE | UIQM | AG |
---|
Average | 0.586 | 4.302 | 5.754 | 0.630 | 4.879 | 11.361 | 1 | 0.586 | 4.484 | 3.180 | 0.610 | 4.970 | 7.970 | 2 | 0.621 | 3.722 | 3.777 | 0.671 | 4.445 | 7.292 | 3 | 0.590 | 4.097 | 7.766 | 0.601 | 4.811 | 13.132 | 4 | 0.609 | 4.028 | 5.470 | 0.623 | 4.524 | 8.997 | 5 | 0.562 | 4.633 | 5.943 | 0.637 | 4.940 | 10.293 | 6 | 0.590 | 4.610 | 5.867 | 0.630 | 4.674 | 11.307 | 7 | 0.638 | 4.894 | 6.947 | 0.642 | 4.834 | 11.302 | 8 | 0.478 | 3.018 | 3.528 | 0.610 | 5.310 | 14.334 | 9 10 | 0.616 0.566 | 4.816 4.716 | 9.540 5.518 | 0.640 0.626 | 4.642 5.642 | 15.520 13.462 |
|
Table 1. Comparison of evaluation indexes of images processed by different algorithms