CN109886901B - Night image enhancement method based on multi-channel decomposition - Google Patents
Night image enhancement method based on multi-channel decomposition Download PDFInfo
- Publication number
- CN109886901B CN109886901B CN201910218341.9A CN201910218341A CN109886901B CN 109886901 B CN109886901 B CN 109886901B CN 201910218341 A CN201910218341 A CN 201910218341A CN 109886901 B CN109886901 B CN 109886901B
- Authority
- CN
- China
- Prior art keywords
- image
- channel
- layer image
- color
- night
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Landscapes
- Image Processing (AREA)
- Facsimile Image Signal Circuits (AREA)
Abstract
The invention discloses a night image enhancement method based on multi-channel decomposition, which is applied to the technical field of image processing and aims at solving the problem that the noise cannot be well removed in the night image processing of the prior art; then, carrying out brightness adaptive calculation on the base layer image; secondly, correcting the brightness of the substrate layer to adapt to the color of the image; performing edge protection and noise suppression on the detail layer image again; finally, fusing the base layer image after color correction and the detail layer image after edge protection and noise suppression to obtain an enhanced night image; the method can well remove the noise interference of the night image.
Description
Technical Field
The invention belongs to the field of image processing, and particularly relates to a night image enhancement technology.
Background
In image processing, images acquired in night scenes often have problems of low brightness, uneven illumination, large noise interference and the like. Poor quality night images tend to affect the performance of computing systems based on image information, such as night monitoring systems and the like. Therefore, preprocessing (such as denoising, brightness enhancement, detail enhancement and the like) for performing visual enhancement on the low-quality night image acquired by the device is of great significance. Typical night or Low brightness image enhancement methods are compared with light source estimation-based image enhancement methods (LIME), see the documents "X.Guo, Y.Li, and H.Ling," LIME: Low-light image enhancement video estimation, "IEEE transactions. image Processing, vol.26, No.2, pp.982-993,2017. The method has good brightness enhancement effect. For the processing of noise, the scheme adopted in the method is that after image enhancement, an image denoising method is additionally added as post-processing. However, since image noise is amplified during the brightness enhancement process, the existing denoising algorithms do not remove noise well.
Disclosure of Invention
In order to solve the technical problem, the invention provides a night image enhancement method based on multi-channel decomposition, which can not amplify noise when the brightness is enhanced.
The technical scheme adopted by the invention is as follows: a nighttime image enhancement method based on multi-pass decomposition comprises the following steps:
s1, decomposing the night image to be processed into a base layer image and a detail layer image according to the estimation of the global noise level of the night image to be processed; the method specifically comprises the following steps: extracting red, green and blue color channels of the night image to be processed, and performing noise estimation on the image of each color channel to obtain an estimated value reflecting the global noise level of each channel; and taking the global noise level estimation value of each channel as a regularization parameter, and decomposing the color channel image based on a total variation model to obtain a substrate layer image and a detail layer image after decomposition of each color channel.
S2, performing brightness adaptation calculation on the base layer image; the method specifically comprises the following steps: extracting a brightness channel of the basal layer image by adopting color space transformation, and then carrying out brightness adaptation calculation on the brightness channel image to obtain a brightness channel image after brightness adaptation:
wherein the content of the first and second substances,
ωg(x,y)＝Lin(x,y)k，ωl(x,y)＝1-ωg(x,y)，n＝exp(σg) (ii) a (x, y) denotes the coordinates of the image pixel points, Lin(x, y) is a luminance channel image extracted from the base layer image, MgIs the pixel mean, S, of the luminance channel imagegIs the standard deviation, S, of the image pixels of the luminance channell(x, y) is the local standard deviation at the location of a pixel point (x, y) in the luminance channel image, Lout(x, y) is the luminance channel image after luminance adaptation, k is the set parameter, wsIs a contrast weighting factor.
S3, performing color correction on the base layer image after brightness adaptation; the method specifically comprises the following steps: combining the luminance channel images before and after luminance adaptation for each color channel with the background layer image of a certain color channel obtained in step S1, performing color correction processing on the background layer image of the color channel to obtain a color-corrected background layer image:
wherein the content of the first and second substances,
three color channels, L, for the original base layer imagein(x, y) is the luminance channel image before luminance adaptation, Lout(x, y) is the adapted brightness channel image, and the value range of the parameter s is as follows: [0,1]。
S4, performing edge protection and noise suppression on the detail layer image; the method specifically comprises the following steps: absolute value processing is carried out on the detail layer image of each color channel to obtain an absolute value image corresponding to each color channel, and Gaussian filtering processing is carried out on the absolute value image corresponding to each color channel to obtain local energy of each pixel point; multiplying the local energy value of each pixel point by the gray value of the pixel point in the corresponding detail layer image in sequence to obtain the value of each pixel point after noise suppression; and finally, obtaining the detail layer image of each color channel after edge protection and noise suppression.
And S5, fusing the base layer image obtained in the step S3 and the detail layer image obtained in the step S4 to obtain an enhanced night image.
The invention has the beneficial effects that: the method comprises the steps of firstly, carrying out noise estimation on three channels of red, green and blue of an image to be processed, and carrying out image decomposition on each channel by taking the set of noise level estimation values as parameters to obtain a basal layer image and a detail layer image; extracting a brightness channel from the base layer image through color space transformation, performing brightness adaptation calculation on the brightness channel, performing color correction on the base layer brightness adaptation image to obtain a brightness adapted base layer image, performing local energy estimation in the detail layer image and realizing self-adaptive noise suppression; finally, carrying out weighted fusion on the base layer image after brightness adaptation and the detail layer image after edge protection and noise suppression to obtain an enhanced night image; the method disclosed by the invention can not amplify noise when the brightness is enhanced through image decomposition, can well remove the noise interference at night through edge protection and noise suppression, and is very suitable for the enhancement processing of low-quality night images.
Drawings
Fig. 1 is a flow chart of a night image enhancement method based on a multi-vision path according to the present invention.
FIG. 2 is a set of contrast maps for enhancing a night image using the method of the present invention in an example;
fig. 2(a) is an original night image, fig. 2(b) is an enhanced image calculated by a contrast method (LIME), and fig. 2(c) is an enhanced image calculated by an image enhancement method of the present invention.
Detailed Description
In order to facilitate the understanding of the technical contents of the present invention by those skilled in the art, the present invention will be further explained with reference to the accompanying drawings.
The invention is further illustrated with reference to the figures and the specific examples.
A nighttime image (number: N-002-0) is selected from a low-quality image library PKU-EAQA (download address: https:// pkuml.org/resources/PKU-eaqa.html) disclosed at present as an implementation object, the image size is 400 x 300 x 3, and the format is 24-bit bmp format image. The flow of the specific calculation method is shown in fig. 1, and the specific process is as follows:
s1, decomposing an image into a base layer and a detail layer based on the estimation of the global noise level of the image: extracting three color channels of red (R), green (G) and blue (B) of an image to be processed, and performing noise estimation on the image of each channel, wherein in this embodiment, the obtained noise level estimation values of the three channels of red, green and blue are respectively: 0.0086, 0.0087, 0.0086; respectively decomposing the image of each channel based on a total variation model by taking the noise level estimation value of each channel as a regularization parameter to obtain a substrate layer image and a detail layer image after decomposition of each channel; taking the pixel (100 ) as an example, the gray values of the pixel in the three channels of red, green and blue in the input image are 0.0784, 0.0745 and 0.0588 respectively. After image decomposition, the gray values of the pixel point in the three channels of the base layer image are respectively 0.0900, 0.0865 and 0.0705, and the values of the pixel point in the three channels of the detail layer image are respectively-0.0116, -0.0120 and-0.0116.
The image decomposition method used in step S1 is a method based on a total variation model, i.e., by solving the following optimization problem:
c belongs to { R, G, B }, and the decomposed image is a basal layer image
And the corresponding detail layer image is
Wherein λcC ∈ { R, G, B } is a regularization parameter, means a gradient operator, and (x, y) means image pixel point coordinates. The estimated value of the global noise level is expressed as:
wherein the content of the first and second substances,
c∈{R,G,B}，Icfor the image to be processed, (x, y) represents the coordinates of the image pixel points,
denotes convolution operation, W denotes the width of the image, and H denotes the height of the image.
S2, performing brightness adaptation calculation on the base layer image; in the present embodiment, the base layer image (RGB image) obtained in step S1 is converted into HSV color space, and the V channel is extracted as a luminance channel. Taking pixel points (100 ) as an example, the values of the point at H, S, V channel after being transformed into HSV space are respectively: 0.136, 0.2169 and 0.0900, extracting a V channel as a brightness channel, namely the value of the pixel point (100 ) in the brightness channel is 0.0900; in this embodiment, the setting value of the parameter k is 0.01, and the parameter w issThe set value is 5.0, and the brightness channel image is brightenedThe value of the brightness channel image after brightness adaptation at the pixel points (100 ) obtained by the adaptive calculation is 0.8148; the specific calculation formula is as follows:
wherein the content of the first and second substances,
ωg(x,y)＝Lin(x,y)k，ωl(x,y)＝1-ωg(x,y)，n＝exp(σg) (ii) a (x, y) denotes the coordinates of the image pixel points, Lin(x, y) is a luminance channel image extracted from the base layer image, MgIs the pixel mean, S, of the luminance channel imagegIs the standard deviation, S, of the image pixels of the luminance channell(x, y) is the local standard deviation at the location of a pixel point (x, y) in the luminance channel image, Lout(x, y) is a brightness channel image after brightness adaptation, k is a set parameter, and the value range is as follows: [0,1]，wsThe value range is the contrast weight coefficient: [0, ∞).
S3, correcting the color of the base layer image; the parameter value s was set to 0.6, and a luminance-adapted base layer image (RGB image) was obtained after color correction. Taking a pixel point (100 ) as an example, the median value of the luminance channel of the base layer image of the point before the original luminance adaptation calculation is 0.0900, the luminance channel image value of the point after the luminance adaptation is 0.8148, and the gray values of the pixel point in the input image in the three channels of red, green and blue are 0.0784, 0.0745 and 0.0588, so after the color correction calculation, the gray values of the pixel point in the three channels of the base layer image after the color correction are: 0.8148, 0.7957, 0.7037; the specific calculation formula is as follows:
wherein the content of the first and second substances,
three color channels, L, for the original base layer imagein(x, y) is the luminance channel image before luminance adaptation, Lout(x, y) is the adapted brightness channel image, and the value range of the parameter s is as follows: [0,1]And e is belonging, representing the relationship between the element and the set, i.e. c is an element in the set R, G, B.
S4, aiming at edge protection and noise suppression of the detail layer image; the method specifically comprises the following steps: obtaining absolute value images by taking absolute values of all channels and all pixel values of the detail layer images, and then carrying out Gaussian filtering processing on all channels of the absolute value images to obtain local energy of each pixel point; for each channel and each pixel point in the image, multiplying the local energy value of each point by the gray value of the point in the detail map in sequence to obtain the value of each pixel point subjected to noise suppression; finally, obtaining a detail layer image subjected to edge protection and noise suppression; taking pixel points (100 ) as an example, the gray values of the point in the three color channels of the detail layer image are respectively: -0.0116, -0.0120, -0.0116, after taking absolute values: 0.0116, 0.0120 and 0.0116; and then, carrying out Gaussian filtering processing on each channel of the absolute value image to obtain the local energy of the pixel points (100 ) in the three channels: 0.6942, 0.6841, 0.5739; multiplying the local energy value of each channel and each point by the gray value of the point in the detail graph to obtain the value of each pixel point after noise suppression, wherein the value is as follows: -0.0080, -0.0082, -0.0067;
s5, fusing the base layer image subjected to brightness adaptation and the detail layer image subjected to edge protection and noise suppression; the value range of the weight coefficient is as follows: [0, ∞), in the present embodiment, the weighting coefficient is set to 5.0; taking the pixel point (100 ) as an example, the value of the pixel point in the luminance-adapted base layer image obtained in step S3 is: 0.8148, 0.7957 and 0.7037, the median value of the detail layer image after the edge protection and the noise suppression obtained in step S4 of the pixel point is-0.0080, -0.0082 and-0.0067, and the product of the median value and the weighting coefficient is: 0.6942, 0.6841, 0.5739; the values of the pixel points (100 ) in the finally obtained enhanced night image are as follows: 0.7748, 0.7547, 0.6703;
the test results are shown in fig. 2, where fig. 2(a) is the original night image, fig. 2(b) is the enhanced image calculated by the LIME of the prior art, and fig. 2(c) is the enhanced image calculated by the image enhancement method of the present invention. As can be clearly seen from fig. 2, when the method of the present invention is used to enhance the night image, a better denoising effect can be obtained, for example, obvious patch-like noise exists in a darker area (ground area) in fig. 2(b), and the ground area in fig. 2(c) is more uniform, i.e., the noise suppression effect of the method of the present invention is more significant.
It will be appreciated by those of ordinary skill in the art that the embodiments described herein are intended to assist the reader in understanding the principles of the invention and are to be construed as being without limitation to such specifically recited embodiments and examples. Various modifications and alterations to this invention will become apparent to those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.
Claims (6)
1. A nighttime image enhancement method based on multi-channel decomposition is characterized by comprising the following steps:
s1, decomposing the night image to be processed into a base layer image and a detail layer image according to the estimation of the global noise level of the night image to be processed; by solving the optimization problem:
obtaining the decomposed image as a base layer image
And the corresponding detail layer image is
Wherein λcC ∈ { R, G, B } is a regularization parameter,
the gradient operator is represented by the expression of the gradient operator,Ic(x, y) represents an image to be processed, and (x, y) represents coordinates of pixel points of the image;
s2, performing brightness adaptation calculation on the base layer image; the method specifically comprises the following steps: extracting a brightness channel of the base layer image by adopting color space transformation, and then performing brightness adaptation calculation on the brightness channel image to obtain a brightness channel image after brightness adaptation;
the expression of the luminance channel image after luminance adaptation is:
wherein the content of the first and second substances,
ωg(x,y)＝Lin(x,y)k，ωl(x,y)＝1-ωg(x,y)，n＝exp(σg) (ii) a (x, y) denotes the coordinates of the image pixel points, Lin(x, y) is a luminance channel image extracted from the base layer image, MgIs the pixel mean, S, of the luminance channel imagegIs the standard deviation, S, of the image pixels of the luminance channell(x, y) is the local standard deviation at the location of a pixel point (x, y) in the luminance channel image, Lout(x, y) is the luminance channel image after luminance adaptation, k is the set parameter, wsIs a contrast weight coefficient;
s3, performing color correction on the base layer image after brightness adaptation;
s4, performing edge protection and noise suppression on the detail layer image;
and S5, fusing the base layer image obtained in the step S3 and the detail layer image obtained in the step S4 to obtain an enhanced night image.
2. The method for enhancing the night image based on the multi-pass decomposition as claimed in claim 1, wherein the step S1 is specifically as follows: extracting red, green and blue color channels of the night image to be processed, and performing noise estimation on the image of each color channel to obtain an estimated value reflecting the global noise level of each channel; and taking the global noise level estimation value of each channel as a regularization parameter, and decomposing the color channel image to obtain a substrate layer image and a detail layer image after decomposition of each color channel.
3. The method of claim 2, wherein the channel image is decomposed based on a fully-variant model.
4. The method for enhancing the night images based on the multi-pass decomposition as claimed in claim 3, wherein the step S3 is specifically as follows: the color correction processing is performed on the background layer image of a certain color channel obtained in step S1 by combining the luminance channel images before luminance adaptation and after luminance adaptation for each color channel, and a color-corrected background layer image is obtained.
5. The method for enhancing night images based on multi-pass decomposition as claimed in claim 4, wherein the expression of the color corrected base layer image is as follows:
wherein L isout(x, y) is the adapted brightness channel image, and the value range of the parameter s is as follows: [0,1]。
6. The method for enhancing the night images based on the multi-pass decomposition as claimed in claim 5, wherein the step S4 is specifically as follows: absolute value processing is carried out on the detail layer image of each color channel to obtain an absolute value image corresponding to each color channel, and Gaussian filtering processing is carried out on the absolute value image corresponding to each color channel to obtain local energy of each pixel point; multiplying the local energy value of each pixel point by the gray value of the pixel point in the corresponding detail layer image in sequence to obtain the value of each pixel point after noise suppression; and finally, obtaining the detail layer image of each color channel after edge protection and noise suppression.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910218341.9A CN109886901B (en) | 2019-03-21 | 2019-03-21 | Night image enhancement method based on multi-channel decomposition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910218341.9A CN109886901B (en) | 2019-03-21 | 2019-03-21 | Night image enhancement method based on multi-channel decomposition |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109886901A CN109886901A (en) | 2019-06-14 |
| CN109886901B true CN109886901B (en) | 2021-07-02 |
Family
ID=66933543
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910218341.9A Active CN109886901B (en) | 2019-03-21 | 2019-03-21 | Night image enhancement method based on multi-channel decomposition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109886901B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023138273A1 (en) * | 2022-01-18 | 2023-07-27 | 武汉联影医疗科技有限公司 | Image enhancement method and system |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102129673A (en) * | 2011-04-19 | 2011-07-20 | 大连理工大学 | Color digital image enhancing and denoising method under random illumination |
| CN103034986A (en) * | 2012-11-29 | 2013-04-10 | 奇瑞汽车股份有限公司 | Night vision image enhancement method based on exposure fusion |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWI273835B (en) * | 2005-07-01 | 2007-02-11 | Ali Corp | Image strengthened system |
| US9240034B2 (en) * | 2012-06-08 | 2016-01-19 | Advanced Micro Devices, Inc. | Biomedical data analysis on heterogeneous platform |
| CN103413275B (en) * | 2013-07-26 | 2016-05-18 | 北京工业大学 | Based on the Retinex nighttime image enhancing method of gradient zero Norm minimum |
| CN106327443A (en) * | 2016-08-24 | 2017-01-11 | 电子科技大学 | Night image enhancement method based on improved genetic algorithm |
-
2019
- 2019-03-21 CN CN201910218341.9A patent/CN109886901B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102129673A (en) * | 2011-04-19 | 2011-07-20 | 大连理工大学 | Color digital image enhancing and denoising method under random illumination |
| CN103034986A (en) * | 2012-11-29 | 2013-04-10 | 奇瑞汽车股份有限公司 | Night vision image enhancement method based on exposure fusion |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109886901A (en) | 2019-06-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104156921B (en) | Self-adaptive low-illuminance or non-uniform-brightness image enhancement method | |
| WO2016206087A1 (en) | Low-illumination image processing method and device | |
| Singh et al. | A novel dehazing model for remote sensing images | |
| CN109816608B (en) | Low-illumination image self-adaptive brightness enhancement method based on noise suppression | |
| Wang et al. | Variational single nighttime image haze removal with a gray haze-line prior | |
| CN111429370A (en) | Method and system for enhancing images in coal mine and computer storage medium | |
| CN107240081A (en) | The denoising of night scene image and enhancing processing method | |
| Xu et al. | Colour image enhancement by virtual histogram approach | |
| WO2018166083A1 (en) | Application of contrast enhancement and visual saliency optimization method in golf course image | |
| CN110009574B (en) | Method for reversely generating high dynamic range image from low dynamic range image | |
| Tsutsui et al. | Halo artifacts reduction method for variational based realtime retinex image enhancement | |
| CN114972102A (en) | Underwater image enhancement method based on global variable contrast enhancement and local correction | |
| CN114998122A (en) | Low-illumination image enhancement method | |
| US9083913B2 (en) | Image processing method, image processing device, and program | |
| CN109886901B (en) | Night image enhancement method based on multi-channel decomposition | |
| WO2020107308A1 (en) | Low-light-level image rapid enhancement method and apparatus based on retinex | |
| CN110545414B (en) | Image sharpening method | |
| CN110992287B (en) | Method for clarifying non-uniform illumination video | |
| JP2015211325A (en) | Purple fringe cancellation processing method and purple fringe cancellation processing device for implementing the processing | |
| CN105741248A (en) | Method for removing hazed and degraded part in image | |
| CN116630198A (en) | Multi-scale fusion underwater image enhancement method combining self-adaptive gamma correction | |
| CN113012079B (en) | Low-brightness vehicle bottom image enhancement method and device and storage medium | |
| Tang et al. | Sky-preserved image dehazing and enhancement for outdoor scenes | |
| CN112465715B (en) | Image scattering removal method based on iterative optimization of atmospheric transmission matrix | |
| CN114170101A (en) | Structural texture keeping low-light image enhancement method and system based on high-frequency and low-frequency information |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |