1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
|
/**
\page tutorial-imgproc-auto-gamma Tutorial: Automatic gamma correction
\tableofcontents
\section imgproc_auto_gamma_intro Introduction
The ViSP library have implementations of different automatic gamma correction methods:
- VISP_NAMESPACE_NAME::GAMMA_LOG_BASED \cite Scott2009TowardsRH
- VISP_NAMESPACE_NAME::GAMMA_NONLINEAR_BASED \cite Shi2007ReducingIB
- VISP_NAMESPACE_NAME::GAMMA_CDF_BASED \cite Huang2013EfficientCE
- VISP_NAMESPACE_NAME::GAMMA_CLASSIFICATION_BASED \cite Rahman2016AnAG
- VISP_NAMESPACE_NAME::GAMMA_SPATIAL_VARIANT_BASED \cite Lee2010ASL
The following example also available in tutorial-compare-auto-gamma.cpp will show how to compare the results of the
different gamma correction methods:
\include tutorial-compare-auto-gamma.cpp
\section imgproc_auto_gamma_cmd_line Program arguments
The main command line arguments can be viewed using the help (`-h`) option, for instance:
- `--input Sample_low_brightness.png` to process a single image or `--input Img_%03d.png` for a sequence of images
- `--max-resolution 600` to reduce the input image resolution and avoid saving too big images
- Gaussian blur options: `--gaussian-kernel-size <e.g. 3, 5, 7> --gaussian-std <e.g. 1>`
- derivative filter aperture size: `--aperture-size <e.g. 3>`
- Canny filtering type: `--canny-filtering-type <0=CANNY_GBLUR_SOBEL_FILTERING, 1=CANNY_GBLUR_SCHARR_FILTERING>`
- color space to be used: `--gamma-rgb (RGB colorspace, else HSV)`
- path to the folder results: `--output <folder path>`
The following image shows the results of the different automatic gamma correction methods for the
"Sample_low_brightness.png" image:
\image html img-tutorial-imgproc-auto-gamma.jpeg
Different metrics are displayed:
- the image entropy, see the [scikit image doc](https://scikit-image.org/docs/dev/auto_examples/filters/plot_entropy.html)
and the [original code](https://github.com/dengyueyun666/Image-Contrast-Enhancement/blob/cd2b1eb5bf6396e2fc3b94cd27f73933d5467147/src/Ying_2017_CAIP.cpp#L186-L207)
- the mean value of the Canny image
- the processing time
The main idea is the better illuminated the image is, the greater is the entropy (the more informative the image is)
and Canny (the more edges can be detected) values are.
\section imgproc_auto_gamma_datasets Low-light image datasets
Some low-ligh datasets are available online, for instance:
- [Exclusively Dark (ExDark) Image Dataset](https://github.com/cs-chan/Exclusively-Dark-Image-Dataset/tree/master/Dataset)
- [Awesome Low Light Image Enhancement](https://github.com/zhihongz/awesome-low-light-image-enhancement)
- [Deep Retinex Decomposition for Low-Light Enhancement](https://daooshee.github.io/BMVC2018website/)
\section imgproc_auto_gamma_next Next tutorial
You can now read the \ref tutorial-imgproc-contrast-sharpening, for additional contrast and sharpness improvement techniques.
*/
|
