CN113096033A - Low-illumination image enhancement method based on Retinex model self-adaptive structure - Google Patents

Low-illumination image enhancement method based on Retinex model self-adaptive structure Download PDF

Info

Publication number
CN113096033A
CN113096033A CN202110304780.9A CN202110304780A CN113096033A CN 113096033 A CN113096033 A CN 113096033A CN 202110304780 A CN202110304780 A CN 202110304780A CN 113096033 A CN113096033 A CN 113096033A
Authority
CN
China
Prior art keywords
image
low
illumination
retinex
map
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.)
Pending
Application number
CN202110304780.9A
Other languages
Chinese (zh)
Inventor
王润雨
齐娜
齐景仲
朱青
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing University of Technology
Original Assignee
Beijing University of Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Beijing University of Technology filed Critical Beijing University of Technology
Priority to CN202110304780.9A priority Critical patent/CN113096033A/en
Publication of CN113096033A publication Critical patent/CN113096033A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G06T5/94—
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
    • G06T5/70—
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/40—Analysis of texture
    • G06T7/49—Analysis of texture based on structural texture description, e.g. using primitives or placement rules
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20004—Adaptive image processing
    • G06T2207/20012—Locally adaptive

Abstract

The invention discloses a low-illumination image enhancement method based on a Retinex model self-adaptive structure, which is improved according to low-illumination conditions and comprises the following steps: a TSRI algorithm for input image illumination structure and reflection texture extraction is proposed. According to the characteristics of the low-illumination image, sparse reflection map L0 is introduced, and the problem that the reflection map in the Retinex algorithm is not smooth enough is solved; aiming at Retinex low-illumination image enhancement algorithm. The TSRI algorithm is used for processing the brightness component in the HSV space.

Description

Low-illumination image enhancement method based on Retinex model self-adaptive structure
Technical Field
The invention relates to a low-illumination image enhancement method based on a Retinex model self-adaptive structure, and belongs to the technical field of image processing.
Background
With the vigorous development of social media such as Facebook and Youtube, some people take or record their daily lives and share them with others through the internet, which is almost a daily meal at home. However, many photographs are taken in low light environments due to backlighting, exposure, or dark environments. Despite improvements in both technology and equipment, the low quality of taking photographs is inevitable, not to mention the large number of photographs that remain a lot of problems in the past.
In low light environments. First, these photographs, i.e., low light images, suffer from low sharpness, low contrast, and significant noise. Post-processing techniques are needed to improve the visual quality of these images. Secondly, the quality of the images and video captured by the optical imaging device may be reduced, which may also reduce the performance of certain systems, such as systems for intelligent traffic analysis, visual surveillance and consumer electronics. Low light conditions in a nighttime environment can produce low contrast images and videos, reducing visibility. Although professional equipment and advanced photography techniques can alleviate these drawbacks to some extent, the inherent factor of noise is inevitable and cannot be addressed at the hardware level. Without a sufficient amount of light, the output of the camera sensor tends to be masked by the inherent noise of the system. While a longer exposure time can effectively improve the signal-to-noise ratio and produce a noise-free image, it raises new problems, such as motion blur. Image enhancement plays an important role in image processing and analysis. The main purpose of image enhancement is to process the image to make it more suitable for certain specific areas than the original image. Nowadays, image enhancement techniques are used in various scientific and engineering fields. Such as atmospheric sciences, astronomy, biomedicine, computer vision, and the like.
Therefore, software-level low-light image enhancement techniques are very popular in photography. Moreover, this technique can also benefit many computer vision algorithms (target detection, tracking, etc.). Since their performance is highly dependent on the clarity of the target scene.
The low-illumination image can be used after being enhanced, and the existing enhancement algorithm has a limit to the improvement of the image quality. The enhancement algorithm based on Retinex theory can simultaneously give consideration to brightness enhancement, detail enhancement and color fidelity, improves the image quality more comprehensively, and is the key point of low-illumination image enhancement research.
In the Retinex model, an image is decomposed into two parts, namely an incident component and a reflection component, wherein the incident component is used for describing information such as a light source and the brightness of a shooting environment, and the reflection component is used for feeding back information on the surface of an object and describing the essential attributes of the object. Solving the incident and reflected components from an image is a problem and an accurate solution cannot be obtained. Some scholars obtain an approximate solution of the incident component through a Gaussian function and image convolution, and further solve a reflection component reflecting the essential attribute of the image to achieve the purpose of enhancement. In order to simplify the calculation amount and the processing convenience, and at the same time, to better conform to the human vision, usually, the logarithm processing is performed on the formula (1), and the multiplication operation is converted into the addition and subtraction operation, i.e. the addition and subtraction operation
Figure BDA0002987171000000021
R (x, y) is the output result of Retinex algorithm. With the continuous development of Retinex algorithm, a plurality of enhanced algorithms based on Retinex theory appear, and representative algorithms include SSR algorithm, MSR algorithm and MSRCR algorithm.
The low-illuminance image generally refers to an image acquired in a shooting environment where light intensity is dark. Typical low-illumination images are classified into two types, one type is that the overall brightness of the image is low; another type of image contains brighter regions, but the overall brightness of the target region of interest is very low.
Disclosure of Invention
The main research content of the method is to provide a low-illumination image enhancement method based on a Retinex model self-adaptive structure. Several common Retinex algorithms are applied to the field of low-illumination image enhancement, and have good effects, and the contrast, the details and the like are improved. However, the conventional Retinex algorithm has certain limitations. Early path-based methods, i.e., the reflection component, could be calculated by multiplying the ratio over some random path. These methods require careful parameter adjustment and can result in high computational costs. The total variation method introduces smooth and reflective assumptions into the variation model to estimate illumination and reflection. However, this method has the side effect of logarithmic transformation that the reflectivity is excessively smoothed. The method firstly provides a new weighting matrix Retinex image enhancement algorithm with a self-adaptive structure, namely a TSRI algorithm for extracting an illumination structure and a reflection texture of an input image. According to the characteristics of low-illumination images, sparse reflection map L0 is introduced, the problem that the reflection map in the Retinex algorithm is not smooth enough is solved, and the effects of adding different weighting matrixes and controlling the iteration times of the weighting matrixes on an output result map are researched.
The structure and texture weighting matrix is proposed:
the Retinex model decomposes the observed scene into its illumination and reflection components. This problem is highly ill-posed, requiring a priori regularization of the solution space taking into account the appropriate illumination and reflectivity. Qualitatively, the illumination should be smooth, capturing the structure of objects in the scene, while the reflectivity should present the physical characteristics of the observed scene, capturing its texture information. Previous structure-texture decomposition methods typically use a TV regularization model to preserve edges. These TV regularization models simply enforce the gradient similarity of the scene and extract the structure of the objects. There are two methods of texture decomposition. One is to derive the structure directly using structure-preserving techniques and the other is to extract the structure from the estimated texture matrix, however, these techniques are prone to ringing. To better understand the role of these techniques in structure/texture extraction, a new filter, i.e., the mean local variance, is proposed.
Figure BDA0002987171000000031
Where Ω is a local block around each pixel of O, whose size was set to 3x3 in all experiments. But the filters presented in the equations cannot be directly applied to a specific problem, so by exponential growth or decay, the extraction of local derivatives will allow extraction of structural and texture information. Adding an index to the filter allows for more flexible separation of structure and texture extraction. As also described by Retinex theory, the larger derivative is due to the change in reflectivity, while the smaller derivative appears in smooth illumination.
Figure BDA0002987171000000032
As also described by Retinex theory, the larger derivative is due to the change in reflectivity, while the smaller derivative appears in smooth illumination. Based on this observation, by setting I0＝R0＝O0.5Put forward
Figure BDA0002987171000000033
Where γ s >1 and γ t <1 are two exponential parameters used to adjust the structural and texture perception of illumination and reflection decomposition.
The structure and texture TSRI objective function is proposed:
a structure and texture aware Retinex model is proposed to simultaneously estimate the illumination I and the reflectivity R of the observed image O. To make the model as simple as possible, the illumination and reflectance components are regularized using the L2 norm.
Figure BDA0002987171000000034
Where S0 and T0 are two matrices defined, the structure should be small enough to preserve the edges of objects in the scene, but large enough to suppress detail, and the texture map should be small enough to show detail. Therefore, adding a sparse representation of L0 to the reflection map yields more accurate results.
The solution of the algorithm is optimized with an alternating direction least squares method:
since the objective function is the separable two variables I and R, the problem can be solved by an alternating direction multiplier algorithm, the two separate sub-problems being convex, ill-posed, and solved alternately. Initializing a matrix variable I0＝R0＝O0.5. Will IkAnd RkAre denoted as kth (k is 0, 1, 2,…) the illumination and reflectance variables at the iteration, L is the number of iterations. By optimizing one variable at a time while fixing the other variable, the two variables can be alternately updated. The method meets the L0 model in the solving process, and can introduce an auxiliary variable h into the reflection mappAnd vpSo that R is also converted to an alternating direction least squares solution. In addition, it is verified whether the weighting matrix should be updated simultaneously when the histogram and the reflection map are updated iteratively.
The TSRI algorithm with the adaptive structure has a good effect on the low-illumination image, but in experiments, it is found that the effect of enhancing the low-illumination image still needs to be improved. When the traditional enhancement algorithm estimates the reflection component of the low-illumination image, the effect of human eyes for perceiving the image is easily reduced, and a great amount of loss of image details is also caused. In order to solve the problem of the traditional operation, when the image is processed, the model improves the effect of the image by changing parameters in the model, and a satisfactory result is obtained.
Drawings
Fig. 1 is a simplified illustration of the image enhancement of the Retinex model.
Fig. 2 is a process flow overview chart.
Fig. 3 is a structural texture extraction diagram.
FIG. 4 is a graph of reflectance comparisons for different iterations of the weighting matrix.
Fig. 5 is a graph comparing the latest low light enhancement algorithm.
FIG. 6 shows the influence of brightness and saturation of the low-illumination image enhancement algorithm based on the RIST model on the visual effect of the image.
Detailed Description
The method firstly provides a Retinex enhancement algorithm of a self-adaptive weighting matrix for the low-illumination image, and the algorithm introduces a new exponential local derivative so as to better utilize the global property of the derivative and popularize the related derivative into structure and texture mapping; respectively extracting illumination and reflection information from the algorithm input image, and simultaneously carrying out L0 sparse expression on the reflection map; the algorithm uses an alternating direction least squares method to solve the ill-posed problem in a better way. The improvement scheme is as follows:
first, a new objective function is applied to the L0 smooth solution, so the method of solving the L0 will be described in detail herein.
On the input map, R is the input image and W is the output result. Gradient of gradient
Figure BDA0002987171000000041
The color difference between adjacent pixels in the x and y directions for each pixel p is calculated. The gradient measure is then:
Figure BDA0002987171000000051
calculation of P as
Figure BDA0002987171000000052
The number of the carbon atoms is not 0. W can be defined as:
Figure BDA0002987171000000053
Wpis the sum of the reflectogram gradients. The above formula also ensures the similarity of image structures. (7) The formula is discrete and difficult to solve, so the alternating direction least square method is also adopted. Introducing an auxiliary variable hpAnd vpRespectively correspond to
Figure BDA0002987171000000054
The target function is retrieved as:
Figure BDA0002987171000000055
C(h,v)＝#{P||hp|+|vp| ≠ 0} and
Figure BDA0002987171000000056
is a parameter to control the gradient of (h, v) and its similarityThe similarity between them. When in use
Figure BDA00029871710000000512
When large enough, (8) is similar to (7). W and (h, v) are solved with an alternating direction least squares method.
a) Solving W: the subproblem minimization (8) for W is:
Figure BDA0002987171000000057
by omitting phases not related to W, equation (9) is obtained, and by deriving it, a minimum solution can be obtained.
b) The objective function for (h, v) is:
Figure BDA0002987171000000058
c (h, v) is the number of non-0 elements in (h, v), then (10) can be written as:
Figure BDA0002987171000000059
because of H (| H)p|+|vpI) is a binary function if H (| H)p|+|vpIf | is 0, then go back to 1, otherwise
Return to 0. Each pixel p in (11) is:
Figure BDA00029871710000000510
minimum E achieved under this conditionPComprises the following steps:
Figure BDA00029871710000000511
by derivation of the above equation, the minimum value E can be obtainedPAnd adding all pixels to obtain an optimal solution.
Figure BDA0002987171000000065
Automatically adjusting in a fixed maximum and minimum value. The algorithm iterates 20 times.
In summary, a solution of L0 of the input map R can be obtained.
The method for solving the L0 is introduced, and the framework mainly comprises the steps of structure texture extraction and illumination component iterative optimization synthesis. A Retinex model is adopted for an input image, a weighting matrix is added, structure and reflection components are extracted, and the components are subjected to reflection structure variation constraint. For the solving step, because the problem is not a proper problem, the least square method is adopted, the reflection component is solved by fixing the illumination component, and the illumination component is solved by fixing the reflection component. And finally, obtaining an output low-light enhancement image.
And re-proposing an optimized objective function according to the proposed weighting matrix:
Figure BDA0002987171000000061
o is the observed image, I represents the scene illumination map representing the brightness of the object, and R represents the surface reflection representing the physical characteristics of the scene. S0And T0Is a weighting matrix that is set forth in terms of the requirement that the structure should be small enough to preserve the edges of objects in the scene, yet large enough to suppress detail, and the texture map should be small enough to show detail. And L0 sparseness is carried out on the reflection map, so that smoothness is guaranteed.
Since the objective function is two variables I and R that are separable, the problem can be solved by an alternating direction multiplier (ADMM) algorithm. Initializing a matrix variable I0＝R0＝O0.5. Will IkAnd RkRespectively, as the illumination and reflectance variables at the k-th (k-0, 1, 2, …) iteration, and L is the number of iterations. By optimizing one variable at a time while repairing the other variable, the two variables can be alternately updated. According to the illumination and reflectivity of each iteration, S can be paired in sequence at the same time0And T0And performing iterative updating. Finally will beThe derivative is set to zero and the solution for illumination and reflectivity is obtained.
And (3) algorithm optimization:
a) update R at fixed I in the kth iteration: : the optimization problem with R becomes:
Figure BDA0002987171000000062
to solve this problem, it is converted into a vector. For this purpose, the vectorization operator vec (·) is used to represent the vector o ═ vec (o), i ═ vec (i), rk＝vec(Rk)，s0＝vec(S0) Their length is nm. Representing the Toeplitz matrix of the discrete gradient operator by G to obtain
Figure BDA0002987171000000066
. By using
Figure BDA0002987171000000063
Figure BDA0002987171000000064
Is represented by rk，s0The matrix located on the main diagonal. Then, the problem is converted to a standard least squares regression problem:
Figure BDA0002987171000000071
the above formula can be converted into:
Figure BDA0002987171000000072
wherein the content of the first and second substances,
Figure BDA0002987171000000073
r 'is'k+1The derivative is set to 0, resulting in the solution:
Figure BDA0002987171000000074
Figure BDA0002987171000000075
regarding | | r | non-conducting phosphor0It can be known that, in the experimental process, it is noted that in the process of extracting the illumination structure, the edge of the illumination pattern is not very significant, and in order to remove a small non-zero gradient, smooth unimportant details and obtain a smoother reflection pattern, the L0 paradigm can be introduced to sparsely express the reflection pattern. Effectively sharpening the main edges by increasing the steepness of the transition while eliminating manageable low-amplitude structures.
Smoothing by L0 on the reflection map R yields:
Figure BDA0002987171000000076
with respect to the smooth solution of the reflection map L0, the method proposed by the foregoing can be adopted.
b) Similarly, I is updated when R is fixed in the kth iteration: : obtaining:
Figure BDA0002987171000000077
the algorithm is as follows:
Figure BDA0002987171000000078
Figure BDA0002987171000000081
and selecting a typical low-illumination image for testing aiming at a TSRI algorithm based on Retinex. The main steps for enhancing the image are as follows:
step 1: inputting a low-illumination image, extracting R, G, B color channels, and converting the color channels into a v channel of an hsv space for experiment;
step 2: adding a weight matrix S0 to the initial map I0 and a weight matrix R0 to the initial map R0;
and step 3: l0 smoothing of the reflection map R0;
and 4, step 4: solving Ik by a conjugate gradient method, solving Rk by the conjugate gradient method, and performing iterative updating;
and 5: updating Sk and Tk for K times in an iteration mode on the basis of Ik and Rk, wherein K is more than or equal to 20;
step 6: carrying out gamma correction on the radiograph;
and 7: obtaining a result illumination map and a result reflection map;
and 8: and outputting a final enhancement result.
Through verification, the enhancement algorithm improves the image quality, enriches the detail information of the image, and shows that the enhancement capability of the algorithm on the detail information of the low-illumination image is strong.
In addition, in order to make the enhancement effect more natural, luminance correction and saturation correction are introduced.
And (3) brightness correction: the brightness of an image is not linear with the brightness perceived by the human eye and is affected by many factors, including physical and psychological factors. The perception of brightness by the human visual system is limited, and the stimulus to vision must reach a certain proportion before it can be perceived by the human eye, and this minimum proportion value is called the human eye recognition threshold. According to weber-fisher's law, the following relationships exist:
Figure BDA0002987171000000082
where Δ I is the change in brightness, I is the actual physical brightness, and the relationship between the brightness perceived by the human eye and the physical brightness of the image can be obtained by integrating both sides of equation (22) simultaneously:
K＝c1·log I+c2 (23)
it can be seen from the expression of formula (23) that the process of human eye perception of brightness is nonlinear, satisfying logarithmic characteristics. Therefore, in adjusting the image brightness, it is necessary to perform nonlinear adjustment. When an image is actually observed, the human eye's perception of brightness is affected by the illumination intensity of an object, and depends on the structure of the image. The brightness of the image background also affects the subjective perception of the viewer. Since the visual characteristics of human eyes are directional, more sensitive to vertical and horizontal directions and less sensitive to brightness changes in other directions, the weights in the vertical and horizontal directions are selectively increased when estimating the background brightness, and the weights in the oblique directions are correspondingly reduced. Therefore, the average background intensity of the pixel points under the RIST model is as follows:
Figure BDA0002987171000000091
wherein alpha and beta are weight coefficients, and Q and D are 4 neighborhoods of the upper, lower, left and right sides of the image pixel point and 4 neighborhoods of the diagonal line respectively.
For the place with higher background brightness, the limitation of brightness enhancement is needed to avoid the brightness unnatural phenomenon of the enhanced image. The result of the luminance adjustment on the V component is as follows:
Figure BDA0002987171000000092
through the adjustment of the above formula, the brightness enhancement problem in the image can be effectively improved, and the enhanced image brightness is more in line with the visual characteristics of human eyes.
And (3) correcting the saturation: in the RGB color space, changing the values of the color channels may cause the image to change significantly, for example, increasing the value of the R component may cause the image to be reddish, and increasing the values of the R and G components may cause the image to be yellowish. In the HSV color space, increasing the image brightness does not shift the image colors to other colors because the brightness and color components are separated. However, the brightness increase of the high-brightness region may cause the saturation to become light and the color of the image to fade, and the specific effect is shown in fig. 6. Fig. (b) shows the color fading phenomenon of the high luminance area of the image after increasing the luminance. The result after the saturation correction is shown in fig. c. The image effect after correction is better than the effect of simply improving the brightness.
After the luminance component is enhanced by using the Retinex algorithm, the internal relationship between the luminance and the saturation of the image changes, which causes the color fading of the enhanced image, so that the saturation component needs to be adjusted to make the saturation change along with the luminance component. The brightness and saturation of the image are not completely independent, and can be cooperatively changed according to the correlation coefficient of the brightness and the saturation. The following formula is therefore used for the adjustment, such as the saturation of the image:
S′(x,y)＝S(x,y)+t(V′(x,y)-ρ(x,y)×V(x,y)) (26)
where ρ (x, y) represents a correlation coefficient of an image, defined as follows:
Figure BDA0002987171000000101
omega is a n x n neighborhood window of the image, inside which window the mean value of the luminance is defined
Figure BDA0002987171000000102
And mean of saturation
Figure BDA0002987171000000103
The following formula:
Figure BDA0002987171000000104
Figure BDA0002987171000000105
wherein the content of the first and second substances,
Figure BDA0002987171000000106
Figure BDA0002987171000000107
the S component is changed along with the change of the V component through the correlation coefficient of the V component and the S component, the saturation is correspondingly changed while the brightness is adjusted, the color fading of the image can be avoided, and the final enhancement result is more natural.
In the description of the present invention, "a plurality" means two or more unless otherwise specified.
It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential attributes thereof; the present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein; any reference sign in a claim should not be construed as limiting the claim concerned.
Furthermore, it should be understood that although the present description refers to embodiments, not every embodiment may contain only a single embodiment, and such description is for clarity only, and those skilled in the art should integrate the description, and the embodiments may be combined as appropriate to form other embodiments understood by those skilled in the art.

Claims (3)

1. The low-illumination image enhancement method based on the Retinex model self-adaptive structure is characterized by comprising the following steps of: the method comprises the following steps:
step 1: inputting a low-illumination image, extracting R, G, B color channels, and converting the color channels into a v channel of an hsv space for experiment;
step 2: adding a weight matrix S0 to the initial map I0 and a weight matrix R0 to the initial map R0;
and step 3: l0 smoothing of the reflection map R0;
and 4, step 4: solving Ik by a conjugate gradient method, solving Rk by the conjugate gradient method, and performing iterative updating;
and 5: updating Sk and Tk for K times in an iteration mode on the basis of Ik and Rk, wherein K is more than or equal to 20;
step 6: carrying out gamma correction on the radiograph;
and 7: obtaining a result illumination map and a result reflection map;
and 8: and outputting a final enhancement result.
2. The low-illumination image enhancement method based on the Retinex model adaptive structure of claim 1, characterized in that: respectively extracting illumination and reflection information from the input image, and simultaneously carrying out L0 sparse expression on the reflection map; an alternating direction least squares method is used.
3. The low-illumination image enhancement method based on the Retinex model adaptive structure of claim 2, characterized in that: aiming at a Retinex low-illumination image enhancement algorithm based on a TSRI model; according to the relation between the human eye visual characteristic and the image background brightness, the enhanced brightness component is subjected to nonlinear adjustment, so that the visual effect of the enhanced image conforms to the human eye visual characteristic; and correcting the saturation by using the correlation characteristics of the brightness and the saturation so as to ensure that the color sense of the image is fuller.
CN202110304780.9A 2021-03-22 2021-03-22 Low-illumination image enhancement method based on Retinex model self-adaptive structure Pending CN113096033A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110304780.9A CN113096033A (en) 2021-03-22 2021-03-22 Low-illumination image enhancement method based on Retinex model self-adaptive structure

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110304780.9A CN113096033A (en) 2021-03-22 2021-03-22 Low-illumination image enhancement method based on Retinex model self-adaptive structure

Publications (1)

Publication Number Publication Date
CN113096033A true CN113096033A (en) 2021-07-09

Family

ID=76668883

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110304780.9A Pending CN113096033A (en) 2021-03-22 2021-03-22 Low-illumination image enhancement method based on Retinex model self-adaptive structure

Country Status (1)

Country Link
CN (1) CN113096033A (en)

Similar Documents

Publication Publication Date Title
Ying et al. A bio-inspired multi-exposure fusion framework for low-light image enhancement
US8503778B2 (en) Enhancing photograph visual quality using texture and contrast data from near infra-red images
Jiang et al. A switched view of Retinex: Deep self-regularized low-light image enhancement
CN107798661B (en) Self-adaptive image enhancement method
CN112734650B (en) Virtual multi-exposure fusion based uneven illumination image enhancement method
KR102261532B1 (en) Method and system for image dehazing using single scale image fusion
Liu et al. Survey of natural image enhancement techniques: Classification, evaluation, challenges, and perspectives
Wang et al. Variational single nighttime image haze removal with a gray haze-line prior
Vazquez-Corral et al. A fast image dehazing method that does not introduce color artifacts
Park et al. Generation of high dynamic range illumination from a single image for the enhancement of undesirably illuminated images
Rahman et al. Efficient image enhancement model for correcting uneven illumination images
CN113039576A (en) Image enhancement system and method
Lepcha et al. A deep journey into image enhancement: A survey of current and emerging trends
Singh et al. Weighted least squares based detail enhanced exposure fusion
He et al. SCENS: Simultaneous contrast enhancement and noise suppression for low-light images
Tang et al. A local flatness based variational approach to retinex
Wang et al. Single Underwater Image Enhancement Based on $ L_ {P} $-Norm Decomposition
Guo et al. A survey on image enhancement for Low-light images
Wei et al. An image fusion dehazing algorithm based on dark channel prior and retinex
Hsieh et al. Variational contrast-saturation enhancement model for effective single image dehazing
Wen et al. Autonomous robot navigation using Retinex algorithm for multiscale image adaptability in low-light environment
Liu et al. Color enhancement using global parameters and local features learning
Wen et al. A survey of image dehazing algorithm based on retinex theory
Zhao et al. Multi-scene image enhancement based on multi-channel illumination estimation
Fan et al. Haze editing with natural transmission

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