Fig. 1. Flow chart of proposed low-illumination enhancement algorithm
Fig. 2. Enhanced images of atmospheric light value via different methods. (a) Original image; (b) method in Ref. [11]; (c) proposed method
Fig. 3. Transmittance images under different α and γ. (a) α=1, γ=0; (b) α=0.75, γ=0.25; (c) α=0.25,γ=0.75; (d) α=0,γ=1
Fig. 4. Edge-extraction images for different bootstrap images. (a) Original image; (b) gray-level inversion image; (c) dark-channel image; (d) illumination images
Fig. 5. Enhancement results after improving transmittance by different methods. (a) Original image; (b) dark-channel; (c) bright-channel; (d) transmittance fusion image; (e) method in Ref. [19]; (f) proposed method
Fig. 6. Enhancement results of low-illumination images with large sky area and obvious change in depth of field for different methods. (a1)--(a4) Original images; (b1)--(b4) method in Ref. [11]; (c1)--(c4) method in Ref. [22]; (d1)--(d4) method in Ref. [23]; (e1)--(e4) proposed method
Fig. 7. Enhancement results of low-illumination images with light source and inhomogeneous light rays for different methods. (a1)--(a3) Original images; (b1)--(b3) method in Ref. [11]; (c1)--(c3) method in Ref. [22]; (d1)--(d3) method in Ref. [23]; (e1)--(e3) proposed method
Bootstrap image category | Similarity |
---|
Gray-level inversion image | 0.1949 | Dark-channel image | 0.3940 | Illumination image | 0.5566 |
|
Table 1. Similarity between different bootstrap images and original images
Original image | Method in Ref. [11] | Method in Ref. [22] | Method in Ref. [23] | Proposed method |
---|
Fig.6(a1) | 7.2516 | 7.6565 | 7.4933 | 7.8762 | Fig.6(a2) | 6.7724 | 7.0452 | 7.1503 | 7.4747 | Fig.6(a3) | 7.5497 | 7.5141 | 7.1526 | 7.7957 | Fig.6(a4) | 6.8558 | 7.5090 | 7.1526 | 7.5188 | Fig.7(a1) | 6.3663 | 6.9588 | 7.0763 | 7.1562 | Fig.7(a2) | 5.5262 | 6.2352 | 6.2431 | 6.7724 | Fig.7(a3) | 6.0362 | 6.7964 | 6.7625 | 7.0236 |
|
Table 2. Information entropies of images
Original image | Method in Ref. [11] | Method in Ref. [22] | Method in Ref. [23] | Proposed method |
---|
Fig.6(a1) | 8.9607 | 11.8851 | 9.7361 | 10.0400 | Fig.6(a2) | 6.8744 | 6.4497 | 5.0609 | 7.9357 | Fig.6(a3) | 4.2377 | 5.3210 | 3.9981 | 6.2678 | Fig.6(a4) | 2.2127 | 2.1871 | 2.8916 | 3.0929 | Fig.7(a1) | 5.2635 | 6.8695 | 6.2302 | 7.0025 | Fig.7(a2) | 3.2163 | 5.6289 | 4.3256 | 5.8953 | Fig.7(a3) | 7.0156 | 7.8635 | 7.2311 | 8.2362 |
|
Table 3. Average gradients of images
Original image | Method in Ref. [11] | Method in Ref. [22] | Method in Ref. [23] | Proposed method |
---|
Fig.6(a1) | 0.5203 | 0.5935 | 0.4988 | 0.6233 | Fig.6(a2) | 0.4032 | 0.5211 | 0.4210 | 0.5834 | Fig.6(a3) | 0.5986 | 0.6525 | 0.5514 | 0.6958 | Fig.6(a4) | 0.5243 | 0.6982 | 0.6201 | 0.7033 | Fig.7(a1) | 0.5722 | 0.5808 | 0.4169 | 0.6110 | Fig.7(a2) | 0.5596 | 0.5117 | 0.4659 | 0.5863 | Fig.7(a3) | 0.4203 | 0.5261 | 0.3034 | 0.5627 |
|
Table 4. Structural similarities of images