CN111640081B - Underwater image recovery method based on optimization and dark channel - Google Patents
Underwater image recovery method based on optimization and dark channel Download PDFInfo
- Publication number
- CN111640081B CN111640081B CN202010517425.5A CN202010517425A CN111640081B CN 111640081 B CN111640081 B CN 111640081B CN 202010517425 A CN202010517425 A CN 202010517425A CN 111640081 B CN111640081 B CN 111640081B
- Authority
- CN
- China
- Prior art keywords
- image
- dark channel
- underwater
- value
- optimization
- 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
- 238000005457 optimization Methods 0.000 title claims abstract description 32
- 238000000034 method Methods 0.000 title claims abstract description 29
- 238000011084 recovery Methods 0.000 title abstract description 7
- 238000013441 quality evaluation Methods 0.000 claims abstract description 19
- 238000011156 evaluation Methods 0.000 claims abstract description 8
- 230000009467 reduction Effects 0.000 claims abstract description 5
- 239000002245 particle Substances 0.000 claims description 44
- 239000013598 vector Substances 0.000 claims description 12
- 230000008569 process Effects 0.000 claims description 7
- 230000015556 catabolic process Effects 0.000 claims description 6
- 238000006731 degradation reaction Methods 0.000 claims description 6
- 238000002834 transmittance Methods 0.000 claims description 6
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 6
- 238000010521 absorption reaction Methods 0.000 claims description 4
- 238000012935 Averaging Methods 0.000 claims description 3
- 238000000149 argon plasma sintering Methods 0.000 claims description 3
- 238000000889 atomisation Methods 0.000 claims description 3
- 239000000284 extract Substances 0.000 claims description 3
- 230000001133 acceleration Effects 0.000 claims description 2
- 230000000694 effects Effects 0.000 abstract description 11
- 230000008859 change Effects 0.000 abstract description 4
- 238000001303 quality assessment method Methods 0.000 abstract 1
- 238000004458 analytical method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 244000144992 flock Species 0.000 description 2
- 230000006399 behavior Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000019637 foraging behavior Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000013535 sea water Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/90—Dynamic range modification of images or parts thereof
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30168—Image quality inspection
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Processing (AREA)
Abstract
The invention discloses an underwater image recovery method based on optimization and a dark channel, which comprises the following steps: (1) dark channel reduction: reducing the underwater image based on classical dark channel theory; (2) no reference image quality assessment: using the non-reference image quality as an evaluation index, estimating the quality of the image according to the self characteristics of the distorted image; (3) overall optimization: and using the non-reference image quality as an evaluation index, and optimizing parameters in the classical dark channel theory by using an optimization algorithm. The invention can automatically change the depth of field restoring parameters of different input underwater images based on classical dark channel theory and non-reference image quality evaluation, and output the images with the best image restoring effect, thereby improving the quality of underwater image restoration.
Description
Technical Field
The invention relates to the field of image processing, in particular to an underwater image recovery method based on optimization and a dark channel.
Background
Under the current situation that the land resource development cannot meet the demands of people, people gradually shift the development targets to ocean resources. When developing ocean resources, people need to use cameras for shooting and analysis, but due to scattering of underwater suspended particles and different absorption effects of seawater on light rays with different wavelengths, image quality is easily distorted and degraded, so that the method cannot be used for further analysis. Therefore, the method is very important for the restoration work of the underwater image and has high engineering application value.
Among the existing numerous image restoration theories, the dark channel theory is a relatively well-known one, and has relatively good restoration effect due to a very ideal physical theory model, so that extensive researches are carried out and a lot of branches are formed. However, this theory has a problem that the parameters used for the reduction are empirically set and are not changed all the time. By using the traversal method, it is found that, for different images, one parameter called depth of field restoration has an important influence on the restoration effect, different depth of field restoration parameters directly influence the output result, and different images have different optimal depth of field parameters, so that the parameters must be different from image to image.
If one method is available, the depth of field restoring parameters of the input different underwater images can be automatically changed based on the dark channel theory, so that the output is carried out with the best image restoring effect, the optimal effect is achieved, and the quality of the underwater image restoration can be greatly improved. Therefore, the above problems need to be solved.
Disclosure of Invention
The invention aims to solve the technical problem of providing an underwater image restoration method based on optimization and dark channels, which can automatically change depth of field restoration parameters of different input underwater images based on classical dark channel theory and non-reference image quality evaluation and output the parameters with the best image restoration effect, thereby improving the quality of underwater image restoration.
In order to solve the technical problems, the invention adopts the following technical scheme: the invention discloses an underwater image recovery method based on an optimization and dark channel, which is characterized by comprising the following steps of:
(1) Dark channel reduction: reducing the underwater image based on classical dark channel theory;
(2) No reference image quality evaluation: using the non-reference image quality as an evaluation index, estimating the quality of the image according to the self characteristics of the distorted image;
in the above step, the no-reference image quality evaluation is a blind image quality prediction BLINDER by a multi-level depth representation, which extracts multi-level representations from a DNN model VGGnet having 37 layers, calculates one feature representation on each layer, then estimates a quality score of each feature vector, and finally estimates the overall quality by averaging these quality scores;
(3) And (3) overall optimization: using the non-reference image quality as an evaluation index, and optimizing parameters in the classical dark channel theory by using an optimization algorithm;
in the above steps, the optimization algorithm adopts a particle swarm algorithm, and the specific flow of the overall optimization combining dark channel restoration and no reference image quality evaluation is as follows:
(3.1) initializing: firstSetting the maximum iteration times IT max The number of independent variables of the objective function is 1, and the maximum speed V of the particles max The location information is the entire search space; setting the particle swarm scale as M, and randomly initializing the speed V in the speed interval and the search space id And position X id Wherein i=1 to M;
(3.2) individual extremum and global optimal solution: defining a fitness function, wherein an individual extremum is an optimal solution found by each particle, finding a global value from the optimal solutions, comparing the global value with historical global optimal values, and updating; (3.3) updating the formula of speed and position as follows:
V id =ωV id +C 1 random(0,1)(P id -X id )+C 2 random(0,1)(P gd -X id )(9)
V id =min(V id ,V max ) (10)
X id =X id +V id (11)
wherein V is id Is the update rate of the position of the particle in space; x is X id Is the position of the particle in space, i.e. the value of the scene depth factor in classical dark channel theory; ω is called the inertia factor; p (P) id D-th dimension, which is the current individual optimum value of the i-th individual variable; p (P) gd D-th dimension, which is the global history optimal value; c (C) 1 And C 2 Is an acceleration factor;
(3.4) judging termination conditions: if one of the two conditions is met, go to step (3.5); otherwise, returning to the step (3.2);
wherein, two conditions are respectively:
(3.4.1) reaching the set iteration count IT max ;
(3.4.2) the global optimum difference between each algebra meets a minimum limit.
(3.5) output: x with the global optimum at this point as the best BLINDER_best id And (3) taking the underwater image as a depth factor parameter w, and outputting a result of underwater image restoration as an optimal result according to a classical dark channel theory.
Preferably, in the step (1), the underwater image is restored by using classical dark channel theory, and the specific process is as follows:
(1.1) a phenomenon that an underwater image is degraded due to problems of light scattering and water absorption, one is a decrease in contrast of the image, and the other is a decrease in contrast; whereas the atomization model of the classical DCP algorithm is:
I(x)=J(x)t(x)+A(1-t(x)) (1)
wherein, I (x) is an underwater distorted image, which is a known image; j (x) is an underwater real image, and is an image to be solved; t (x) is transmittance; a is water background estimation;
degradation of the underwater image and degradation of the image in air are similar;
(1.2) defining a dark channel according to the following formula;
J dark (x)=min y∈Ω(x) (min c∈(r,g,b) J c (y)) (2)
wherein J is dark (x) Representing the dark channel image as a single channel image and having a scalar value; x= [ m, n] T Representing pixel coordinate vectors in the image, m and n being pixel coordinate values; j (J) c (y) represents an image of each channel in the original image, and its value is a scalar; c represents three channels of the image red, green and blue, and Ω (x) represents a window centered on pixel x; y= [ m, n ]] T Representing a pixel coordinate vector in a widget Ω (x);
(1.3) for any one of the input images, taking its dark channel image gray value J dark (x) Maximum 0.1% pixel point, average value of gray value of each channel corresponding to pixel position of original input image, and calculating atmospheric light value A of each channel c ;
(1.4) A obtained according to the formula (1) and the above step (1.3) c The formula is obtained:
wherein I is c (x) And J c (x) C-channels respectively representing the known image and the image to be solved;
(1.5) setting the transmittance to a constant valueAnd the two sides of the formula (3) take the minimum value, so as to obtain the formula:
wherein y represents a pixel in a small window omega (x) and is used for distinguishing the pixel from the original x of the whole sub-graph;
(1.6) according to the dark channel prior theory, the dark channel image is approximately 0, i.e
(1.7) substituting the formula (5) into the formula (4) to obtain the formula:
(1.8) in order to prevent defogging from being too thorough, the restored scenery is unnatural, so that a depth factor parameter w is introduced to obtain the formula:
wherein w is 0 to 1;
(1.9) performing image restoration according to the formula (8);
wherein t is 0 To prevent the transmissivity from being too small, resulting in an enhanced image that is too bright; max (t (x), t) 0 ) The largest value among the values in brackets is taken.
Preferably, in the above step (1.3), the atmospheric light value A c Is a three-element vector and each element corresponds to each color channel.
Preferably, in the step (1.8), w is 0.95.
Preferably, in the step (3.2), the fitness function is a BLINDER index in the reference-free image quality evaluation, and the larger the requirement is, the better; the specific process is as follows:
(3.2.1) the current position X of each particle id As depth factor parameter w in each dark channel restoration, then carrying out image restoration on the original image, and carrying out non-reference image quality evaluation on each restored image to obtain BLINDER index of each restored image;
(3.2.2) comparing according to the larger and better criteria to obtain the current individual optimal values of the M particles in the round, and the global historical optimal values of all particles in the whole computation history, and the corresponding spatial positions P thereof id And P gd These spatial positions are essentially the values of the depth factor.
Preferably, in the above step (3.3), V id By a maximum velocity V with the particles max The minimum value is defined as the maximum value; x is X id Is set in the entire search space for the position information.
The invention has the beneficial effects that: the invention can automatically change the depth of field restoring parameters of different input underwater images based on classical dark channel theory and non-reference image quality evaluation, and output the images with the best image restoring effect, thereby improving the quality of underwater image restoration.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments described in the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of an underwater image restoration method based on optimization and dark channels.
Detailed Description
The technical scheme of the present invention will be clearly and completely described in the following detailed description.
The invention discloses an underwater image recovery method based on an optimization and dark channel, which comprises the following steps:
(1) Dark channel reduction: reducing the underwater image based on classical dark channel theory;
in the above steps, the underwater image is restored by using classical dark channel theory, and the specific flow is as follows:
(1.1) a phenomenon that an underwater image is degraded due to problems of light scattering and water absorption, one is a decrease in contrast of the image, and the other is a decrease in contrast; whereas the atomization model of the classical DCP algorithm is:
I(x)=J(x)t(x)+A(1-t(x)) (1)
wherein, I (x) is an underwater distorted image, which is a known image; j (x) is an underwater real image, and is an image to be solved; t (x) is transmittance; a is water background estimation;
degradation of the underwater image and degradation of the image in air are similar;
(1.2) defining a dark channel according to the following formula;
J dark (x)=min y∈Ω(x) (min c∈(r,g,b) J c (y)) (2)
wherein J is dark (x) Representing the dark channel image as a single channel image and having a scalar value; x= [ m, n] T Representing pixel coordinate vectors in the image, m and n being pixel coordinate values; j (J) c (y) represents an image of each channel in the original image, and its value is a scalar; c represents three channels of the image red, green and blue, and Ω (x) represents a window centered on pixel x; y= [ m, n ]] T Representing a pixel coordinate vector in a widget Ω (x);
(1.3) for any one ofAn input image takes its dark channel image gray value J dark (x) Maximum 0.1% pixel point, average value of gray value of each channel corresponding to pixel position of original input image, and calculating atmospheric light value A of each channel c ;
In the above step, the atmospheric light value A c Is a three-element vector and each element corresponds to each color channel.
(1.4) A obtained according to the formula (1) and the above step (1.3) c The formula is obtained:
wherein I is c (x) And J c (x) C-channels respectively representing the known image and the image to be solved;
(1.5) setting the transmittance to a constant valueAnd the two sides of the formula (3) take the minimum value, so as to obtain the formula:
wherein y represents a pixel in a small window omega (x) and is used for distinguishing the pixel from the original x of the whole sub-graph;
(1.6) according to the dark channel prior theory, the dark channel image is approximately 0, i.e
(1.7) substituting the formula (5) into the formula (4) to obtain the formula:
(1.8) in order to prevent defogging from being too thorough, the restored scenery is unnatural, so that a depth factor parameter w is introduced to obtain the formula:
the method for underwater image is characterized in that the depth of field restoration parameters have important influence on the restoration effect between 0 and 1 according to different images, the output result is directly influenced by different depth of field restoration parameters, and different images have different optimal depth of field parameters, so that the parameters are required to be different from image to image;
in the classical dark channel theory, w is 0.95 for a common foggy image;
(1.9) performing image restoration according to the formula (8);
wherein t is 0 To prevent the transmissivity from being too small, resulting in an enhanced image that is too bright; max (t (x), t) 0 ) The largest value among the values in brackets is taken.
(2) No reference image quality evaluation: using the non-reference image quality as an evaluation index, estimating the quality of the image according to the self characteristics of the distorted image;
in the above step, the no-reference image quality evaluation is a blind image quality prediction BLINDER by a multi-level depth representation, which extracts multi-level representations from a DNN model VGGnet having 37 layers, calculates one feature representation on each layer, then estimates a quality score for each feature vector, and finally estimates the overall quality by averaging these quality scores. The method is recognized as a method module for evaluating the quality of the reference-free image because the accuracy is high; the content of which is known and therefore not described in detail herein.
(3) And (3) overall optimization: and the non-reference image quality is used as an evaluation index, and the parameters in the classical dark channel theory are optimized by using an optimization algorithm, so that the recovery effect of the parameters on the underwater image is improved.
Among the many optimization algorithms, particle swarm algorithm (Particle Swarm Optimization, PSO) is a very classical type, which was first proposed by Eberhart and Kennedy in 1995, whose basic concept stems from the study of the foraging behavior of the flock. Consider one such scenario: a group of birds searches randomly for food, only a piece of food in this area, and all birds do not know where the food is, but they know how far from the food they are at the current location. The simplest and most effective strategy is to find the individual in the flock closest to the food to search. The PSO algorithm is inspired from this biological population behavior and used to solve the optimization problem.
The bird individuals are simulated by using particles, each particle can be regarded as a search individual in an N-dimensional search space, the current position of the particle is a candidate solution of a corresponding optimization problem, the flying process of the particle is the search process of the individual, the flying speed of the particle can be dynamically adjusted according to the optimal position of the particle history and the optimal position of the population history, and the particle has only two attributes: speed, which represents the speed of movement, and position, which represents the direction of movement. The optimal solution searched by each particle is called an individual extremum, and the optimal individual extremum in the particle swarm is used as the current global optimal solution. The iteration is continued, updating the speed and the position. And finally obtaining the optimal solution meeting the termination condition.
The optimization algorithm in the invention adopts a particle swarm algorithm, and the specific flow of the overall optimization combining dark channel restoration and no reference image quality evaluation is as follows:
(3.1) initializing: first, a maximum iteration number IT is set max The number of arguments of the objective function is 1 (because there is only one depth factor parameter w), the maximum velocity V of the particle max The position information is the whole search space (w ranges from 0 to 1); setting the particle swarm scale as M, and randomly initializing the speed V in the speed interval and the search space id And position X id Wherein i=1 to M;
in the present embodiment, IT max Set to 100, the number of objective function self-scalar amounts is 1, the maximum velocity V of the particle max The particle size M was 10, which was 0.002.
(3.2) individual extremum and global optimal solution: defining a fitness function, wherein an individual extremum is an optimal solution found by each particle, finding a global value from the optimal solutions, comparing the global value with historical global optimal values, and updating;
in the step, the fitness function is BLINDER index in the non-reference image quality evaluation, and the larger the requirement is, the better the requirement is; the specific process is as follows:
(3.2.1) the current position X of each particle id As depth factor parameter w in each dark channel restoration, then carrying out image restoration on the original image, and carrying out non-reference image quality evaluation on each restored image to obtain BLINDER index of each restored image;
(3.2.2) comparing according to the larger and better criteria to obtain the current individual optimal values of the M particles in the round, and the global historical optimal values of all particles in the whole computation history, and the corresponding spatial positions P thereof id And P gd These spatial positions are essentially the values of the depth factor.
(3.3) updating the formula of speed and position as follows:
V id =ωV id +C 1 random(0,1)(P id -X id )+C 2 random(0,1)(P gd -X id )(9)
V id =min(V id ,V max ) (10)
X id =X id +V id (11)
wherein V is id Is the update rate of the position of the particle in space, is the velocity V in equation (10) by the velocity V at the maximum velocity with the particle max The minimum value is defined as the maximum value; x is X id Is the position of the particle in space, i.e. the value of the depth of view factor in classical dark channel theory, the range of which is set in the whole search space (range of w is 0-1) of the position informationThe method comprises the steps of carrying out a first treatment on the surface of the Omega is called an inertia factor, when the value is large, the global optimizing capability is strong, and the local optimizing capability is weak; when the value is smaller, the global optimizing capability is weakened, and the local optimizing capability is strengthened; p (P) id D-th dimension, which is the current individual optimum value of the i-th individual variable; p (P) gd D-th dimension, which is the global history optimal value; c (C) 1 And C 2 To accelerate the factor, an increasing randomness of balancing the individual optimum with the global historical optimum is achieved, thereby increasing the diversity of the variation.
(3.4) judging termination conditions: if one of the following two conditions is met, go to step (3.5); otherwise, returning to the step (3.2);
in the above steps, two conditions are respectively:
(3.4.1) reaching the set iteration count IT max ;
(3.4.2) the global optimum difference between each algebra meets a minimum limit.
(3.5) output: x with the global optimum at this point as the best BLINDER_best id And (3) taking the underwater image as a depth factor parameter w, and outputting a result of underwater image restoration as an optimal result according to a classical dark channel theory.
The invention has the beneficial effects that: the invention can automatically change the depth of field restoring parameters of different input underwater images based on classical dark channel theory and non-reference image quality evaluation, and output the images with the best image restoring effect, thereby improving the quality of underwater image restoration.
The above embodiments are merely illustrative of the preferred embodiments of the present invention, and the present invention is not limited to the above embodiments, and various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the design concept of the present invention should fall within the protection scope of the present invention, and the claimed technical content of the present invention is fully described in the claims.
Claims (6)
1. An underwater image restoration method based on optimization and dark channels is characterized by comprising the following steps:
(1) Dark channel reduction: reducing the underwater image based on classical dark channel theory;
(2) No reference image quality evaluation: using the non-reference image quality as an evaluation index, estimating the quality of the image according to the self characteristics of the distorted image;
in the above step, the no-reference image quality evaluation is a blind image quality prediction BLINDER by a multi-level depth representation, which extracts multi-level representations from a DNN model VGGnet having 37 layers, calculates one feature representation on each layer, then estimates a quality score of each feature vector, and finally estimates the overall quality by averaging these quality scores;
(3) And (3) overall optimization: using the non-reference image quality as an evaluation index, and optimizing parameters in the classical dark channel theory by using an optimization algorithm;
in the above steps, the optimization algorithm adopts a particle swarm algorithm, and the specific flow of the overall optimization combining dark channel restoration and no reference image quality evaluation is as follows:
(3.1) initializing: first, a maximum iteration number IT is set max The number of independent variables of the objective function is 1, and the maximum speed V of the particles max The location information is the entire search space; setting the particle swarm scale as M, and randomly initializing the speed V in the speed interval and the search space id And position X id Wherein i=1 to M;
(3.2) individual extremum and global optimal solution: defining a fitness function, wherein an individual extremum is an optimal solution found by each particle, finding a global value from the optimal solutions, comparing the global value with historical global optimal values, and updating;
(3.3) updating the formula of speed and position as follows:
V id =ωV id +C 1 random(0,1)(P id -X id )+C 2 random(0,1)(P gd -X id ) (9)
V id =min(V id ,V max ) (10)
X id =X id +V id (11)
wherein V is id Is the update rate of the position of the particle in space; x is X id Is the position of the particle in space, i.e. the value of the scene depth factor in classical dark channel theory; ω is called the inertia factor; p (P) id D-th dimension, which is the current individual optimum value of the i-th individual variable; p (P) gd D-th dimension, which is the global history optimal value; c (C) 1 And C 2 Is an acceleration factor;
(3.4) judging termination conditions: if one of the two conditions is met, go to step (3.5); otherwise, returning to the step (3.2);
wherein, two conditions are respectively:
(3.4.1) reaching the set iteration count IT max ;
(3.4.2) the global optimum difference between each algebra meets a minimum limit;
(3.5) output: x with the global optimum at this point as the best BLINDER_best id And (3) taking the underwater image as a depth factor parameter w, and outputting a result of underwater image restoration as an optimal result according to a classical dark channel theory.
2. The underwater image restoration method based on the optimization and dark channel as claimed in claim 1, wherein: in the step (1), the underwater image is restored by using classical dark channel theory, and the specific process is as follows:
(1.1) a phenomenon that an underwater image is degraded due to problems of light scattering and water absorption, one is a decrease in contrast of the image, and the other is a decrease in contrast; whereas the atomization model of the classical DCP algorithm is:
I(x)=J(x)t(x)+A(1-t(x)) (1)
wherein, I (x) is an underwater distorted image, which is a known image; j (x) is an underwater real image, and is an image to be solved; t (x) is transmittance; a is water background estimation;
degradation of the underwater image and degradation of the image in air are similar;
(1.2) defining a dark channel according to the following formula;
J dark (x)=min y∈Ω(x) (min c∈(r,g,b) J c (y)) (2)
wherein J is dark (x) Representing the dark channel image as a single channel image and having a scalar value; x= [ m, n] T Representing pixel coordinate vectors in the image, m and n being pixel coordinate values; j (J) c (y) represents an image of each channel in the original image, and its value is a scalar; c represents three channels of the image red, green and blue, and Ω (x) represents a window centered on pixel x; y= [ m, n ]] T Representing a pixel coordinate vector in a widget Ω (x);
(1.3) for any one of the input images, taking its dark channel image gray value J dark (x) Maximum 0.1% pixel point, average value of gray value of each channel corresponding to pixel position of original input image, and calculating atmospheric light value A of each channel c ;
(1.4) A obtained according to the formula (1) and the above step (1.3) c The formula is obtained:
wherein I is c (x) And J c (x) C-channels respectively representing the known image and the image to be solved;
(1.5) setting the transmittance to a constant valueAnd the two sides of the formula (3) take the minimum value, so as to obtain the formula:
wherein y represents a pixel in a small window omega (x) and is used for distinguishing the pixel from the original x of the whole sub-graph;
(1.6) according to the dark channel prior theory, the dark channel image is approximately 0, i.e
(1.7) substituting the formula (5) into the formula (4) to obtain the formula:
(1.8) in order to prevent defogging from being too thorough, the restored scenery is unnatural, so that a depth factor parameter w is introduced to obtain the formula:
wherein w is 0 to 1;
(1.9) performing image restoration according to the formula (8);
wherein t is 0 To prevent the transmissivity from being too small, resulting in an enhanced image that is too bright; max (t (x), t) 0 ) The largest value among the values in brackets is taken.
3. The underwater image restoration method based on the optimization and dark channel as claimed in claim 2, wherein: in the above step (1.3), the atmospheric light value A c Is a three-element vector and each element corresponds to each color channel.
4. The underwater image restoration method based on the optimization and dark channel as claimed in claim 2, wherein: in the above step (1.8), w is 0.95.
5. The underwater image restoration method based on the optimization and dark channel as claimed in claim 1, wherein: in the step (3.2), the fitness function is a BLINDER index in the non-reference image quality evaluation, and the larger the requirement is, the better the requirement is; the specific process is as follows:
(3.2.1) the current position X of each particle id As depth factor parameter w in each dark channel restoration, then carrying out image restoration on the original image, and carrying out non-reference image quality evaluation on each restored image to obtain BLINDER index of each restored image;
(3.2.2) comparing according to the larger and better criteria to obtain the current individual optimal values of the M particles in the round, and the global historical optimal values of all particles in the whole computation history, and the corresponding spatial positions P thereof id And P gd These spatial positions are essentially the values of the depth factor.
6. The underwater image restoration method based on the optimization and dark channel as claimed in claim 1, wherein: in the above step (3.3), V id By a maximum velocity V with the particles max The minimum value is defined as the maximum value; x is X id Is set in the entire search space for the position information.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010517425.5A CN111640081B (en) | 2020-06-09 | 2020-06-09 | Underwater image recovery method based on optimization and dark channel |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010517425.5A CN111640081B (en) | 2020-06-09 | 2020-06-09 | Underwater image recovery method based on optimization and dark channel |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111640081A CN111640081A (en) | 2020-09-08 |
CN111640081B true CN111640081B (en) | 2023-04-28 |
Family
ID=72331449
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010517425.5A Active CN111640081B (en) | 2020-06-09 | 2020-06-09 | Underwater image recovery method based on optimization and dark channel |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111640081B (en) |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102411774B (en) * | 2011-08-08 | 2013-05-01 | 安科智慧城市技术(中国)有限公司 | Processing method, device and system based on single-image defogging |
US9087385B2 (en) * | 2012-11-12 | 2015-07-21 | FMV Innovations, LLC. | Method for improving images captured underwater |
CN108734670B (en) * | 2017-04-20 | 2021-05-18 | 天津工业大学 | Method for restoring single night weak-illumination haze image |
CN108961206B (en) * | 2018-04-20 | 2021-05-11 | 北京航空航天大学 | Non-reference objective evaluation method for defogging effect |
CN111028200B (en) * | 2019-11-13 | 2023-05-23 | 南京信息工程大学 | Image optimization method based on non-reference image quality evaluation and MSR |
-
2020
- 2020-06-09 CN CN202010517425.5A patent/CN111640081B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN111640081A (en) | 2020-09-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Zhou et al. | Underwater image restoration via backscatter pixel prior and color compensation | |
CN111310862B (en) | Image enhancement-based deep neural network license plate positioning method in complex environment | |
CN112614077B (en) | Unsupervised low-illumination image enhancement method based on generation countermeasure network | |
CN108596853B (en) | Underwater image enhancement method based on background light statistical model and transmission map optimization | |
CN111241925A (en) | Face quality evaluation method, system, electronic equipment and readable storage medium | |
US20040239762A1 (en) | Adaptive background image updating | |
CN110288550B (en) | Single-image defogging method for generating countermeasure network based on priori knowledge guiding condition | |
CN110246151B (en) | Underwater robot target tracking method based on deep learning and monocular vision | |
CN112115963A (en) | Method for generating unbiased deep learning model based on transfer learning | |
CN113111979B (en) | Model training method, image detection method and detection device | |
Ye et al. | An adaptive image enhancement technique by combining cuckoo search and particle swarm optimization algorithm | |
CN111652822A (en) | Single image shadow removing method and system based on generation countermeasure network | |
CN105447825A (en) | Image defogging method and system | |
CN111274964B (en) | Detection method for analyzing water surface pollutants based on visual saliency of unmanned aerial vehicle | |
CN111310609A (en) | Video target detection method based on time sequence information and local feature similarity | |
Jin et al. | Broad colorization | |
CN115880720A (en) | Non-labeling scene self-adaptive human body posture and shape estimation method based on confidence degree sharing | |
CN111640081B (en) | Underwater image recovery method based on optimization and dark channel | |
Li et al. | A low-light image enhancement method with brightness balance and detail preservation | |
Babu et al. | ABF de-hazing algorithm based on deep learning CNN for single I-Haze detection | |
CN115018729B (en) | Content-oriented white box image enhancement method | |
CN114387484B (en) | Improved mask wearing detection method and system based on yolov4 | |
Wang et al. | Airlight estimation based on distant region segmentation | |
CN111640082B (en) | Underwater image recovery method based on Gaussian mixture model and dark channel theory | |
CN108416815A (en) | Assay method, equipment and the computer readable storage medium of air light value |
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 |