CN106504222B - A kind of underwater Polarization Image Fusion system based on bionic visual mechanism - Google Patents
A kind of underwater Polarization Image Fusion system based on bionic visual mechanism Download PDFInfo
- Publication number
- CN106504222B CN106504222B CN201611047282.6A CN201611047282A CN106504222B CN 106504222 B CN106504222 B CN 106504222B CN 201611047282 A CN201611047282 A CN 201611047282A CN 106504222 B CN106504222 B CN 106504222B
- Authority
- CN
- China
- Prior art keywords
- image
- polarization
- underwater
- fdct
- fusion
- 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
- 230000010287 polarization Effects 0.000 title claims abstract description 211
- 230000004927 fusion Effects 0.000 title claims abstract description 64
- 230000000007 visual effect Effects 0.000 title claims abstract description 41
- 230000007246 mechanism Effects 0.000 title claims abstract description 35
- 239000011664 nicotinic acid Substances 0.000 title claims abstract description 23
- 238000003384 imaging method Methods 0.000 claims abstract description 21
- 244000304921 Charybdis maritima Species 0.000 claims abstract description 18
- LSMIOFMZNVEEBR-ICLSSMQGSA-N scilliroside Chemical compound C=1([C@@H]2[C@@]3(C)CC[C@H]4[C@@]([C@]3(CC2)O)(O)C[C@H](C2=C[C@@H](O[C@H]3[C@@H]([C@@H](O)[C@H](O)[C@@H](CO)O3)O)CC[C@@]24C)OC(=O)C)C=CC(=O)OC=1 LSMIOFMZNVEEBR-ICLSSMQGSA-N 0.000 claims abstract description 18
- 238000005286 illumination Methods 0.000 claims abstract description 15
- 230000008447 perception Effects 0.000 claims abstract description 12
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims abstract description 9
- 230000006870 function Effects 0.000 claims description 39
- 230000009021 linear effect Effects 0.000 claims description 35
- 239000011159 matrix material Substances 0.000 claims description 30
- 230000000873 masking effect Effects 0.000 claims description 29
- 239000005557 antagonist Substances 0.000 claims description 25
- 238000000034 method Methods 0.000 claims description 24
- 230000009466 transformation Effects 0.000 claims description 23
- 230000008485 antagonism Effects 0.000 claims description 21
- 238000004422 calculation algorithm Methods 0.000 claims description 21
- 238000012545 processing Methods 0.000 claims description 19
- 238000007500 overflow downdraw method Methods 0.000 claims description 17
- 238000013507 mapping Methods 0.000 claims description 12
- 230000015572 biosynthetic process Effects 0.000 claims description 11
- 238000003786 synthesis reaction Methods 0.000 claims description 11
- 238000004364 calculation method Methods 0.000 claims description 9
- 238000002156 mixing Methods 0.000 claims description 8
- 238000005457 optimization Methods 0.000 claims description 8
- 239000002245 particle Substances 0.000 claims description 8
- 239000013598 vector Substances 0.000 claims description 8
- 230000003044 adaptive effect Effects 0.000 claims description 7
- 238000002474 experimental method Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- 238000013441 quality evaluation Methods 0.000 claims description 3
- 238000000354 decomposition reaction Methods 0.000 claims description 2
- 235000013399 edible fruits Nutrition 0.000 claims 1
- 238000001514 detection method Methods 0.000 abstract description 9
- 238000004458 analytical method Methods 0.000 abstract description 4
- 230000008901 benefit Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 230000008859 change Effects 0.000 description 4
- 238000013461 design Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000010365 information processing Effects 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 210000004556 brain Anatomy 0.000 description 2
- 238000012512 characterization method Methods 0.000 description 2
- 150000001875 compounds Chemical class 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 239000000155 melt Substances 0.000 description 2
- 238000003672 processing method Methods 0.000 description 2
- 241001465754 Metazoa Species 0.000 description 1
- 238000010521 absorption reaction Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000005290 field theory Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000009022 nonlinear effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000002194 synthesizing effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration by the use of more than one image, e.g. averaging, subtraction
-
- G06T5/70—
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20004—Adaptive image processing
-
- 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
- G06T2207/20212—Image combination
- G06T2207/20221—Image fusion; Image merging
Abstract
The underwater Polarization Image Fusion system based on bionic visual mechanism that the invention discloses a kind of comprising the underwater polarization imaging module of concentrated light, polarization parameter image computing module, image storage module, image grading Fusion Module, output display module based on nonuniform illumination.Firstly, establishing nonuniform illumination as underwater lighting using bundling light source, imitates squill vision polarization mechanism of perception and extract underwater polarization image, inhibit back scattering caused by suspended particulates in water;Then, the feature and Multiscale Fusion of polarization image are realized based on the level based adjustment model of human-eye visual characteristic building image using the information correlativity and complementarity between polarization parameter image.The present invention can effectively improve the clarity and contrast of underwater polarization image, is conducive to the detection and analysis of submarine target, further increases the detection accuracy of submarine target.
Description
Technical field
It is specifically a kind of based on bionic visual mechanism the invention belongs to Underwater Imaging and technical field of image processing
Underwater Polarization Image Fusion system and method.
Background technique
In recent years, with the growth of underwater environment perception demand, underwater optics imaging technique is widely used in underwater mesh
It marks in the fields such as detection, hydraulic engineering, ocean geography engineering investigation and ocean military affairs.However, water body during Underwater Imaging
Scattering and sink effect bring non-linear effects to light, cause underwater picture quality to be remarkably decreased, image detail obscures, makes an uproar
Sound brings great difficulty to the artificial interpretation of image and automatic interpretation compared with strong, contrast is low.Therefore, how effectively to press down
Back scattering processed is the key that improve underwater imaging system performance, has obtained the extensive concern of scholars, non-uniform lighting light field
Theoretical, polarization imaging technology is wherein representative solution.Non-uniform lighting light field theory is divided from light source and light field
Analysis angle reduces the influence of back scattering noise: in the short range of image receiving system, being illuminated with low energy densities, to the greatest extent
The influence of shot noise may be reduced;It for target farther out, is illuminated with high-energy density, improves the intensity of echo signal, realized
Non-blind area receives and improves the clarity of image.Stablizing due to the polarization characteristic variation of light can be predicted, and polarization imaging is avoiding light
Line scattering and absorption etc. have unique advantage, are that new developing direction is imaged in underwater optics.How several are made full use of
The complementarity of polarization image information studies the use processing method of underwater polarization image, improves the quality of image, be allowed to
It is the important research direction of current underwater polarization image processing conducive to subsequent target detection and analysis.
Under water in optical environment, predictable, the polarization of target object rear orientation light is stablized in the polarization characteristic variation of light
Degree is greater than the degree of polarization of suspended particulates rear orientation light, therefore, places linear polarizer or circuit polarizer before the detectors, can subtract
The influence of small suspended particulates rear orientation light, to improve the clarity of Underwater Imaging.Polarization Detection technology is used for underwater mesh
Mark detection, the target acquisition for complicated underwater environment provide a completely new technological approaches.2005, Schechner et al.
It proposes to improve the underwater picture visibility under the conditions of natural illumination by polarization imaging.By processing polarizing film horizontal and vertical
The two images acquired under angle overcome valance effect caused by back scattering, improve scene contrast.2006,
Treibitz et al. is proposed to use wide visual field polarized illumination, analyzer is placed before receiver and two width polarization states of acquisition are mutual
Vertical underwater polarization image, this method can effectively eliminate back scattering effect.Dubreuil et al. utilizes polarization state phase
Mutually orthogonal underwater polarization image and its relativity measurement improves the detection performance of submarine target.
Image co-registration is to be registrated, synthesized to the multiple images of same target, makes full use of the information between multiple images
Complementarity increases the information content of image to overcome the limitation of single image, keeps target image more complete, to improve figure
The reliability and clarity of picture achieve the purpose that enhance picture quality.Since polarization image processing substantially needs to utilize difference
The complementarity of polarization angle image information, it is extensive that the polarization image processing technique based on image interfusion method has obtained researchers
Concern.Consider in complicated underwater environment, studies polarization image use processing method, improve the clear of polarization image
Degree and contrast realize that accurately and reliably Underwater Target Detection is most important to polarization image quality is improved.Polarization Image Fusion
Method is most at present to be applied in ground and remote sensing polarization imaging.Fusion method mainly uses both of which, and one is based on puppet
The image co-registration of color mapped, as Zhao Y et al. proposes to be based on degree of polarization linear modulation and RGB face in ground target classification
The image interfusion method of color model calculates the shared information of polarization parameter image and the angle of polarization, obtained image is mapped to RGB
Color space obtains preliminary pseudo-colours fusion results, then by defining the degree of polarization index of modulation, amendment obtains finally polarizing fusion
Image, to improve the validity of image object classification.Another kind is the image co-registration based on multi-scale transform, such as Zhang Dexiang
People decomposes polarization parameter image and degree of polarization using Directionlet transformation for remote sensing target identification application, right
Obtained low frequency sub-band is first carried out two dimension Teager filtering to high-frequency sub-band and calculated, obtained side using weighted average blending algorithm
Edge and non-edge information, are respectively adopted edges of regions maximum and the maximum strategy of Direction Contrast obtains high frequency blending image, most
Fusion results image is obtained by inverse transformation afterwards.But directly each polarization parameter image is merged, computation complexity is higher;
Only polarization intensity image is selected to merge with degree of polarization, will cause partial polarization information and lose and influence fusion results.
As the marine organisms squill of lower animal, its compound eye possesses superb polarization sensing capability but its brain knot
Structure is simple, and the mankind as superior being, though the sensing capability of human eye is far away from the compound eye of squill, and it is multiple due to having
Miscellaneous brain structure, the vision system of people possess powerful information processing capability.Therefore, the present invention uses for reference the polarization view of squill
Feel perception mechanism, the powerful information processing capability of simulation human visual system proposes a kind of water based on bionic visual mechanism
Lower Polarization Image Fusion system and method.The system, as underwater lighting, is completed imitative using the bundling light source of nonuniform illumination
The underwater polarization imaging and polarization parameter image of squill vision calculate, then realize based on the underwater of human-eye visual characteristic (HVS)
Polarization image level based adjustment, including two levels of Fusion Features and Multiscale Fusion, thus improve polarization image clarity and
Contrast improves the visual effect of image, provides image that is intuitive, clear, being suitable for analysis, patent retrieval and to each both at home and abroad
The newest retrieval of kind scientific and technical literature shows that there has been no correlation techniques to be seen in document.
Summary of the invention
The present invention proposes a kind of water based on bionic visual mechanism to solve defect and deficiency existing in the prior art
The polarization image acquisition and image co-registration processing of submarine target may be implemented in lower Polarization Image Fusion system and method, thus
Obtain the underwater polarization image that detailed information is abundant and contrast is high.
In order to solve the above-mentioned technical problem, the technical scheme adopted by the invention is that:
A kind of underwater Polarization Image Fusion system based on bionic visual mechanism, including underwater polarization imaging module, polarization
Parametric image computing module, image storage module, image grading Fusion Module and output display module;It is described to polarize under water
As module, nonuniform illumination is established as underwater lighting using bundling light source, disposes linear polarizer before video camera under water, led to
Cross polarizing film Rotation Controllers and change polarization angle, acquire respectively polarization angle be 0 °, 45 °, 90 °, 135 ° of four width linear polarizations
Image I0°,I45°,I90°,I135°;
The polarization parameter image computing module simulates the polarization perception mechanism of squill, establishes binary channels polarization antagonism
Model, by the mutually perpendicular linear polarization image I of the four width polarization angle0°,I90°,I45°,I135°Antagonism binary channels is inputted respectively,
By Optimized model parameter, the polarization parameter for antagonist for obtaining and there is different polarization characteristic information, including polarization parameter for antagonist are calculated
Ph and Pd, wherein Ph is indicated by I0°And I90°The output in 1 channel of composition polarizes parameter for antagonist, and Pd is indicated by I45°And I135°Composition
2 channels output polarize parameter for antagonist, and using Stokes vector method calculate linear polarization degree DoP and synthesis light intensity I, thus
The underwater polarization parameter image of four width for obtaining polarization parameter for antagonist Ph, Pd and linear polarization degree DoP, synthesizing light intensity I;
Described image memory module stores acquisition and the underwater polarization parameter image data calculated;
Described image level based adjustment module constructs the level based adjustment model of underwater polarization image based on human-eye visual characteristic,
The Multiscale Fusion for successively realizing first order Fusion Features and the second level obtains the polarization that detailed information is abundant and contrast is high and melts
Close image;
The polarization fusion results image is exported and is shown by the output display module.
A kind of underwater Polarization Image Fusion system based on bionic visual mechanism above-mentioned, it is characterised in that: the polarization
Parametric image computing module imitates the polarization parameter for antagonist calculation method of squill polarization vision, comprising the following steps:
(1) the polarization antagonism for using for reference squill perceives mechanism, establishes binary channels polarization antagonism model, forms two linear polarizations
Antagonism channel, each channel is by mutually orthogonal a pair of of polarization signal as input, i.e., 1 channel is by I0°And I90°As input, 2
Channel is by I45°And I135°As input;
(2) the tuner parameters s and t for designing linear polarization antagonism channel, to the input polarization signal in 1 channel and 2 channels to point
Not carry out linear weighted function processing, obtain twin-channel output polarization parameter for antagonist Ph and Pd.
Wherein, in the step (2), the output polarization parameter for antagonist calculation formula in 1 channel and 2 channels is as follows:
Ph=s × I0°+t×I90°
Pd=s × I45°+t×I135°,
Wherein, s and t is the tuner parameters in linear polarization antagonism channel.
A kind of underwater Polarization Image Fusion system based on bionic visual mechanism above-mentioned, it is characterised in that: described image
In level based adjustment module, the underwater polarization parameter image level based adjustment method based on human-eye visual characteristic, be divided into Fusion Features and
Two levels of Multiscale Fusion, first order Fusion Features the following steps are included:
(1) tri- width polarization parameter image matrix of Ph, Pd and DoP is joined end to end by column respectively and is saved as 3 column vectors, then group
Input data matrix A of the matrix arranged at one 3 as Non-negative Matrix Factorization (NMF)i×3(wherein, i=nm is each image
Sum of all pixels, n, m respectively indicate the row, column number of each image matrix);
(2) A is usedi×3Euclidean distance square is used as objective function between W*H, and the columns of selected characteristic basic matrix W is
1, to Ai×3Matrix carries out NMF processing, to obtain feature bases Wi×1, it is special to get polarization to convert thereof into n × m dimension matrix
Levying blending image PF, W and H is to A respectivelyi×3The feature basic matrix and A of linear approximationi×3Projection coefficient square on feature base
Battle array.
A kind of underwater Polarization Image Fusion system based on bionic visual mechanism above-mentioned, it is characterised in that: described is inclined
Shake parametric image level based adjustment module in, second level Multiscale Fusion method the following steps are included:
(1) fast discrete Curvelet change is carried out to synthesis intensity image I to be fused and polarization characteristic blending image PF
(FDCT) is changed, low frequency sub-band CI is respectively obtained0(x, y) and CPF0(x, y), with each high frequency direction subband CI(s,d)(x, y) and
CPF(s,d)(x, y), wherein s indicates that scale, d indicate subband direction.
(2) to the low frequency sub-band in the step (1), using following linear weighted function convergence strategy:
CFI0(x, y)=w1CI0(x,y)+w2CPF0(x,y)
Wherein, CFI0(x, y) is fused low frequency sub-band, w1、w2For merge weight factor, value range be [0,
1], by the comentropy E (w of fused low-frequency image at this time1,w2) fitness function as Chaos particle swarm optimization algorithm, benefit
Best fusion weight factor w is adaptively found with Chaos particle swarm optimization algorithm1And w2。
(3) in order to retain the detailed information of image to be fused, to the high frequency direction subband in the step (1), using office
Convergence strategy of portion's region energy as characteristic quantity:
Wherein, CFI(s,d)(x, y) is the high-frequency sub-band coefficient in the direction fused s scale d,WithRespectively indicate high-frequency sub-band of the synthesis intensity image I and polarization characteristic image PF on the direction s scale d with (x,
Y) energy of local area centered on, calculation formula are as follows:
In formula: M, N indicate the length and width of the local rectangular portions centered on (x, y);
(4) illumination in image is removed not using Retinex algorithm to fused low frequency sub-band coefficient in step (2)
, noise is filtered out using Adaptive Thresholding to fused high frequency direction sub-band coefficients in step (3);
Wherein, Retinex algorithm is handled using multi-Scale Retinex Algorithm, Retinex algorithm:
R (x, y) is output low frequency subband, and F (x, y) is Gaussian filter function, and S (x, y) is input low frequency sub-band, and K is ruler
The number of degree, w are the weights of each scale.Preferably, K=3, and take w1=w2=w3=1/3,
Adaptive Wavelet Thrinkage method: here, the threshold value of each high frequency direction subband, which is chosen, uses following formula:
In formula, to most thin scale, k=4 is taken, remaining subband takes k=3, the noise criteria difference σ of original imagenWith in robust
Value estimation is estimated, it may be assumed that σn=median (abs (C))/0.6745, high frequency on the direction s scale d
Band parameter varianceIt is calculated using the Monte Carlo estimation technique;
(5) transformation of HVS mask is carried out to treated in step (4) low frequency sub-band coefficient and high frequency direction sub-band coefficients,
Obtain the FDCT contrast masking coefficient based on human-eye visual characteristic;
(6) the FDCT contrast masking coefficient that the step (5) obtains is located automatically using nonlinear mapping function
Reason, the minutia of prominent image;
(7) FDCT low frequency sub-band coefficient is automatically adjusted using nonlinear gain function, improves image overall contrast
Degree;
(8) anti-to HVS mask is carried out through nonlinear mapping function treated FDCT contrast masking coefficient in step (6)
Transformation, the FDCT high frequency direction sub-band coefficients that obtain that treated;
(9) FDCT is carried out to treated in step (7) and (8) FDCT low frequency sub-band coefficient and high frequency direction sub-band coefficients
Inverse transformation obtains final fusion results image FI.
A kind of underwater Polarization Image Fusion system based on bionic visual mechanism above-mentioned, it is characterised in that: described is inclined
Vibration parametric image second level Multiscale Fusion method, it is characterised in that: the HVS mask transformation in the step (5), including brightness
Two step of mask and contrast masking, first carries out luminance mask, the FDCT luminance mask coefficient after obtaining luminance mask are as follows:
Then degree of comparing mask, the FDCT contrast masking coefficient obtained after contrast masking are as follows:
In above-mentioned luminance mask and contrast masking formula, x(1,1)For the low frequency sub-band coefficient of FDCT, x(s,d)For FDCT's
High frequency direction sub-band coefficients on s level, the direction d, s are level, and d is subband direction, as s > 1, indicate high frequency
Band.
A kind of underwater Polarization Image Fusion system based on bionic visual mechanism above-mentioned, it is characterised in that: described is inclined
It shakes parametric image second level Multiscale Fusion method, in the step (6), the FDCT that is obtained after nonlinear mapping function is handled
Contrast masking coefficient are as follows:
Wherein,CLCM(s,d)FDCT before indicating nonlinear mapping function processing
Contrast masking coefficient.
Polarization parameter image second level Multiscale Fusion method above-mentioned, it is characterised in that: in the step (7), through non-
Linear gain function treated FDCT low frequency sub-band coefficient are as follows:
Wherein,x(1,1)For the FDCT low frequency sub-band before nonlinear gain function processing
Coefficient.
A kind of underwater Polarization Image Fusion system based on bionic visual mechanism above-mentioned, it is characterised in that: described is inclined
Shake parametric image second level Multiscale Fusion method, the HVS mask inverse transformation in the step (8) include anti-contrast masking and
Anti- two step of luminance mask, first obtaining FDCT luminance mask coefficient through anti-contrast mask process is
Obtaining FDCT high frequency direction sub-band coefficients through anti-luminance mask again is
A kind of underwater Polarization Image Fusion system based on bionic visual mechanism above-mentioned, it is characterised in that: described is inclined
Vibration parametric image second level Multiscale Fusion method, it is characterised in that: in the step (1), underwater polarization image is carried out
FDCT transformation, taking the level of transformation is ceil (log2(min (M, N)) -3) layer, wherein M, N respectively indicate the height and width of image.
Present invention advantageous effects achieved:
By using above-mentioned technical proposal, it is an advantage of the invention that marine organisms are regarded using bio-information processing means
Feel that the information processing capability of powerful image perception function and human eye vision complexity combines, designs based on bionic visual mechanism
Underwater Polarization Image Fusion system.Start with from the mechanism and feature of underwater optics polarization imaging, uses for reference the polarization vision of squill
Perception mechanism builds underwater polarization experiment platform, design polarization antagonism model, to obtain the underwater polarization image of high quality.Root
According to the difference of characteristic information contained by different polarization parametric image, level based adjustment model is constructed: first with the sparse of human eye vision
Characterization carries out the Fusion Features of polarization parameter image;Then Retinex theory and the visual mask for using for reference human eye vision are special
Property, and multiple dimensioned, multidirectional " sparse " characterization of Curvelet transformation is combined, it is realized in Curvelet transform domain underwater
The Multiscale Fusion of polarization image, to obtain the underwater polarization image that details is clear, contrast is high.Therefore, the present invention can be with
To realize that Underwater Target Detection provides clear reliable image data.
Detailed description of the invention
Connection relationship between configuration and each configuration of the Fig. 1 based on the underwater Polarization Image Fusion system of bionic visual mechanism
Schematic diagram;
The underwater polarization imaging schematic diagram of concentrated light of the Fig. 2 based on nonuniform illumination;
Fig. 3 imitates the polarization parameter for antagonist extraction scheme schematic diagram of squill polarization vision perception;
The general frame of underwater polarization parameter image level based adjustment scheme of the Fig. 4 based on human-eye visual characteristic;
The Multiscale Fusion method flow schematic diagram of the underwater polarization image of Fig. 5.
Specific embodiment
It is existing in order to which auditor can be best understood from technical characteristic of the invention, technology contents and its technical effect reached
Attached drawing of the invention is described in detail in conjunction with the embodiments.However, shown attached drawing, is intended merely to that this hair is better described
Bright technical solution, so, ask auditor not limit claims of the invention with regard to attached drawing.
The invention patent is further illustrated with reference to the accompanying drawing.
As shown in Figure 1, the present invention provides a kind of underwater Polarization Image Fusion system based on bionic visual mechanism, system packet
Include underwater polarization imaging module, polarization parameter image computing module, image storage module, image grading Fusion Module and output
Display module.
The underwater polarization imaging module is established nonuniform illumination as underwater lighting using bundling light source, is taken the photograph under water
Linear polarizer is disposed before the ccd image receiver of camera, polarization angle is changed by polarizing film Rotation Controllers, is acquired respectively
Polarization angle is 0 °, 45 °, 90 °, 135 ° of four width linear polarization image I0°,I45°,I90°,I135°.Fig. 2 is based on nonuniform illumination
The underwater polarization imaging schematic diagram of concentrated light.The a wide range of light beam generated by bundling light source is by underwater reflecting mirror to hydrospace
It is illuminated, the reception optical axis of ccd image receiver and the transmitting optical axis of bundling light source are orthogonal (mutually vertical with two in figure
Straight black dotted lines indicate).Because the angular brightness of light source is exponentially distributed, the light field in entire irradiation area is strong
Degree also exponentially changes, and will obtain low energy densities light in short distance point, and will obtain at distant object point high
Energy density light.Underwater Imaging is carried out under the irradiation of this non-uniform lighting light field, due to being closely that low energy densities are shone
It is bright, therefore back-scattered light is very low, can achieve the lowest threshold of detector, is at a distance then high-energy density lighting area,
Neighbouring scattering light is quite faint after long distance transmission, can omit and disregard, and this reception scheme and the visual angle depth of field
Size is unrelated, it can be achieved that non-blind area receives.
The polarization parameter image computing module simulates the polarization perception mechanism of squill, establishes binary channels polarization antagonism
Model, by the mutually perpendicular linear polarization image I of the four width polarization angle0°,I90°,I45°,I135°Antagonism binary channels is inputted respectively,
By Optimized model parameter, the polarization parameter for antagonist for obtaining and there is different polarization characteristic information, including polarization parameter for antagonist are calculated
Ph and Pd (are indicated with Ph by I0°And I90°The output in 1 channel of composition polarizes parameter for antagonist, and Pd is indicated by I45°And I135°Composition
The output in 2 channels polarizes parameter for antagonist).The polarization parameter for antagonist extraction scheme of imitative squill polarization vision perception is as shown in Figure 3.
Then, using Stokes vector method calculate linear polarization degree DoP and synthesis light intensity I, thus obtain polarization parameter for antagonist Ph, Pd and
The underwater polarization parameter images of four width such as linear polarization degree DoP, synthesis light intensity I.
Described image memory module stores acquisition and the underwater polarization parameter image data calculated;
Described image level based adjustment module constructs the level based adjustment model of underwater polarization image based on human-eye visual characteristic,
The Multiscale Fusion for successively realizing first order Fusion Features and the second level obtains the polarization that detailed information is abundant and contrast is high and melts
Close image;
The polarization fusion results image is exported and is shown by the output display module.
In the polarization parameter image computing module, the polarization parameter for antagonist calculation method of squill polarization vision is imitated,
The following steps are included:
(1) the polarization antagonism for using for reference squill perceives mechanism, establishes binary channels polarization antagonism model, forms two linear polarizations
Antagonism channel, each channel is by mutually orthogonal a pair of of polarization signal as input, i.e., 1 channel is by I0°And I90°As input, 2
Channel is by I45°And I135°As input;
(2) the input polarization signal in 1 channel and 2 channels is obtained twin-channel defeated to linear weighted function processing is carried out respectively
Parameter for antagonist Ph and Pd are polarized out.
Wherein, in the step (2), the output polarization parameter for antagonist calculation formula in 1 channel and 2 channels is as follows:
Ph=s × I0°+t×I90°
Pd=s × I45°+t×I135°
In formula, s and t are the tuner parameters in linear polarization antagonism channel, are determined by experiment the value range of parameter s and t, so
It is excellent using Chaos-Particle Swarm Optimization afterwards using the no-reference image quality evaluation index Q based on local structure tensor as objective function
Change algorithm, adaptively finds the optimal solution of parameter s and t.
Image quality evaluation factor Q based on local structure tensor are as follows:
In formula, s1And s2Singular value decomposition is carried out by partial gradient covariance matrix (local structure tensor) C of image
(SVD) it obtains:
N × N neighborhood of a certain pixel f (x, y) in image is (w), the local gradient vectors J of the point are as follows:
J=[gx(k) gy(k)],k∈w
As shown in figure 4, in described image level based adjustment module, the underwater polarization parameter image based on human-eye visual characteristic
Level based adjustment method is divided into two levels of Fusion Features and Multiscale Fusion, first order Fusion Features the following steps are included:
(1) tri- width polarization parameter image matrix of Ph, Pd and DoP is joined end to end by column respectively and is saved as 3 column vectors, then group
Input data matrix A of the matrix arranged at one 3 as Non-negative Matrix Factorization (NMF)i×3(wherein, i=nm is each image
Sum of all pixels, n, m respectively indicate the row, column number of each image matrix);
(2) A is usedi×3Euclidean distance square is used as objective function between W*H, and the columns of selected characteristic basic matrix W is
1, to Ai×3Matrix carries out NMF processing, to obtain feature bases Wi×1, it is special to get polarization to convert thereof into n × m dimension matrix
Levying blending image PF, W and H is to A respectivelyi×3The feature basic matrix and A of linear approximationi×3Projection coefficient square on feature base
Battle array.
As shown in figure 5, in described image level based adjustment module, second level Multiscale Fusion method the following steps are included:
(1) fast discrete Curvelet change is carried out to synthesis intensity image I to be fused and polarization characteristic blending image PF
(FDCT) is changed, low frequency sub-band CI is respectively obtained0(x, y) and CPF0(x, y), with each high frequency direction subband CI(s,d)(x, y) and
CPF(s,d)(x, y) (wherein s indicates that scale, d indicate subband direction).
(2) to the low frequency sub-band in the step (1), using following linear weighted function convergence strategy:
CFI0(x, y)=w1CI0(x,y)+w2CPF0(x,y)
Wherein, CFI0(x, y) is fused low frequency sub-band, w1、w2For merge weight factor, value range be [0,
1].By the comentropy E (w of fused low-frequency image at this time1,w2) fitness function as Chaos particle swarm optimization algorithm, benefit
Best fusion weight factor w is adaptively found with Chaos particle swarm optimization algorithm1And w2。
(3) in order to retain the detailed information of image to be fused, to the high frequency direction subband in the step (1), using office
The great convergence strategy of portion's region energy:
Wherein, CFI(s,d)(x, y) is the high-frequency sub-band coefficient in the direction fused s scale d,WithRespectively indicate high-frequency sub-band of the synthesis intensity image I and polarization characteristic image PF on the direction s scale d with (x,
Y) energy of local area centered on, calculation formula are as follows:
In formula: M, N indicate the length and width of the local rectangular portions centered on (x, y);
(4) illumination in image is removed not using Retinex algorithm to fused low frequency sub-band coefficient in step (2)
, noise is filtered out using Adaptive Thresholding to fused high frequency direction sub-band coefficients in step (3);
Wherein, Retinex algorithm is handled using multi-Scale Retinex Algorithm, Retinex algorithm:
R (x, y) is output low frequency subband, and F (x, y) is Gaussian filter function, and S (x, y) is input low frequency sub-band, and K is ruler
The number of degree, w are the weights of each scale.Preferably, K=3, and take w1=w2=w3=1/3,
Adaptive Wavelet Thrinkage method: here, the threshold value of each high frequency direction subband, which is chosen, uses following formula:
In formula, to most thin scale, k=4 is taken, remaining subband takes k=3, the noise criteria difference σ of original imagenWith the intermediate value of robust
Estimation is estimated, it may be assumed that σn=median (abs (C))/0.6745, the high-frequency sub-band parameter variance on the direction s scale dIt is calculated using the Monte Carlo estimation technique;
(5) transformation of HVS mask is carried out to treated in step (4) low frequency sub-band coefficient and high frequency direction sub-band coefficients,
Obtain the FDCT contrast masking coefficient based on human-eye visual characteristic;
(6) the FDCT contrast masking coefficient that the step (5) obtains is located automatically using nonlinear mapping function
Reason, the minutia of prominent image;
(7) FDCT low frequency sub-band coefficient is automatically adjusted using nonlinear gain function, improves image overall contrast
Degree;
(8) anti-to HVS mask is carried out through nonlinear mapping function treated FDCT contrast masking coefficient in step (6)
Transformation, the FDCT high frequency direction sub-band coefficients that obtain that treated;
(9) FDCT is carried out to treated in step (7) and (8) FDCT low frequency sub-band coefficient and high frequency direction sub-band coefficients
Inverse transformation obtains final fusion results image FI.
Wherein, the HVS mask transformation in the step (5), including luminance mask and contrast masking
Two steps first carry out luminance mask, the FDCT luminance mask coefficient after obtaining luminance mask are as follows:
Then degree of comparing mask, the FDCT contrast masking coefficient obtained after contrast masking are as follows:
In above-mentioned luminance mask and contrast masking formula, x(1,1)For the low frequency sub-band coefficient of FDCT, x(s,d)For FDCT's
High frequency direction sub-band coefficients on s level, the direction d, s are level, and d is subband direction, as s > 1, indicate high frequency
Band.
In the step (6), the FDCT contrast masking coefficient that is obtained after nonlinear mapping function is handled are as follows:
Wherein,CLCM(s,d)FDCT before indicating nonlinear mapping function processing
Contrast masking coefficient.
In the step (7), through nonlinear gain function treated FDCT low frequency sub-band coefficient are as follows:
Wherein,x(1,1)For the FDCT low frequency sub-band before nonlinear gain function processing
Coefficient.
HVS mask inverse transformation in the step (8) includes two step of anti-contrast masking and anti-luminance mask, first through opposing
Obtaining FDCT luminance mask coefficient than degree mask process is
Obtaining FDCT high frequency direction sub-band coefficients through anti-luminance mask again is
Preferably, in the step (1), FDCT transformation is carried out to underwater polarization image, taking the level of transformation is ceil
(log2(min (M, N)) -3) layer, wherein M, N respectively indicate the height and width of image.
The present invention designs the underwater polarization imaging scheme of concentrated light based on nonuniform illumination, uses for reference biology perception and letter
Treatment mechanism is ceased, the polarization vision mechanism of perception of squill is imitated, construction binary channels polarizes antagonism model, obtains the polarization of high quality
Parametric image, and based on human-eye visual characteristic using the polarization image level based adjustment of two levels of Fusion Features and Multiscale Fusion
Method realizes the adaptive fusion of underwater polarization image, can effectively enhance the edge details feature of image, improves uneven illumination,
Improve contrast and clarity.
Basic principles and main features and advantage of the invention have been shown and described above.The technical staff of the industry should
Understand, the present invention is not limited to the above embodiments, and the above embodiments and description only describe originals of the invention
Reason, without departing from the spirit and scope of the present invention, various changes and improvements may be made to the invention, these changes and improvements
It all fall within the protetion scope of the claimed invention.The claimed scope of the invention is by appended claims and its equivalent
Boundary.
Claims (6)
1. a kind of underwater Polarization Image Fusion system based on bionic visual mechanism, it is characterised in that: including underwater polarization imaging
Module, polarization parameter image computing module, image storage module, image grading Fusion Module and output display module;It is described
Underwater polarization imaging module, establishes nonuniform illumination as underwater lighting using bundling light source, disposes before video camera under water
Linear polarizer, changes polarization angle by polarizing film Rotation Controllers, and acquiring polarization angle respectively is 0 °, 45 °, 90 °, 135 °
Four width linear polarization image I0°,I45°,I90°,I135°;
The polarization parameter image computing module simulates the polarization perception mechanism of squill, establishes binary channels polarization antagonism model,
By the mutually perpendicular linear polarization image I of the four width polarization angle0°,I90°,I45°,I135°Antagonism binary channels is inputted respectively, is passed through
Optimized model parameter, calculate obtain have different polarization characteristic information polarization parameter for antagonist, including polarization parameter for antagonist Ph and
Pd, wherein Ph is indicated by I0°And I90°The output in 1 channel of composition polarizes parameter for antagonist, and Pd is indicated by I45°And I135°The 2 of composition
The output in channel polarizes parameter for antagonist, and calculates linear polarization degree DoP and synthesis light intensity I using Stokes vector method, thus
To the underwater polarization parameter image of four width of polarization parameter for antagonist Ph, Pd and linear polarization degree DoP, synthesis light intensity I;
Described image memory module stores acquisition and the underwater polarization parameter image data calculated;
Described image level based adjustment module constructs the level based adjustment model of underwater polarization image based on human-eye visual characteristic, successively
The Multiscale Fusion for realizing first order Fusion Features and the second level obtains the polarization fusion knot that detailed information is abundant and contrast is high
Fruit image;
The polarization fusion results image is exported and is shown by the output display module;
The polarization parameter image computing module imitates the polarization parameter for antagonist calculation method of squill polarization vision, including following
Step:
(1) the polarization antagonism for using for reference squill perceives mechanism, establishes binary channels polarization antagonism model, it is short of money to form two linear polarizations
Anti- channel, each channel is by mutually orthogonal a pair of of polarization signal as input, i.e., 1 channel is by I0°And I90°It is logical as input, 2
Road is by I45°And I135°As input;
(2) the tuner parameters e and t for designing linear polarization antagonism channel, to the input polarization signal in 1 channel and 2 channels to respectively into
The processing of row linear weighted function obtains twin-channel output polarization parameter for antagonist Ph and Pd,
Wherein, in the step (2), the output tuner parameters calculation formula in 1 channel and 2 channels is as follows:
Ph=e × I0°+t×I90°
Pd=e × I45°+t×I135°,
Wherein, e and t is the tuner parameters in linear polarization antagonism channel;
It is determined by experiment the value range of parameter e and t, is then commented with the no-reference image quality based on local structure tensor
Valence index Q is as objective function, using Chaos particle swarm optimization algorithm, adaptively finds the optimal solution of parameter e and t,
No-reference image quality evaluation index Q based on local structure tensor are as follows:
E and t carries out singular value decomposition by the partial gradient covariance matrix local structure tensor F of image and obtains:
P × P neighborhood of a certain pixel f (x, y) in image is w, the local gradient vectors J of the pixel are as follows:
J=[gx(k) gy(k)],k∈w;
In described image level based adjustment module, the underwater polarization parameter image level based adjustment method based on human-eye visual characteristic, point
Be characterized two levels of fusion and Multiscale Fusion, first order Fusion Features the following steps are included:
(a) tri- width polarization parameter image matrix of Ph, Pd and DoP is joined end to end by column respectively and is saved as 3 column vectors, recomposition one
Input data matrix A of the matrix of a 3 column as Non-negative Matrix Factorizationi×3, wherein i=C*D is the sum of all pixels of each image,
C, D respectively indicates the row, column number of each image matrix;
(b) A is usedi×3Between W × H Euclidean distance square be used as objective function, and the columns of selected characteristic basic matrix W be 1,
To Ai×3Matrix carries out NMF processing, to obtain feature bases Wi×1, C × D dimension matrix is converted thereof into get polarization characteristic
Blending image PF, W and H are to A respectivelyi×3The feature basic matrix and A of linear approximationi×3Projection coefficient matrix on feature base;
In the image grading Fusion Module, second level Multiscale Fusion method the following steps are included:
(1) fast discrete Curvelet transformation is carried out to synthesis intensity image I to be fused and polarization characteristic blending image PF, point
Low frequency sub-band CI is not obtained0(x, y) and CPF0(x, y), with each high frequency direction subband CI(s,d)(x, y) and CPF(s,d)(x, y),
Wherein s indicates that scale, d indicate subband direction;
(2) to the low frequency sub-band in the step (1), using following linear weighted function convergence strategy:
CFI0(x, y)=w1CI0(x,y)+w2CPF0(x,y)
Wherein, CFI0(x, y) is fused low frequency sub-band, w1、w2To merge weight factor, value range is [0,1], will
Comentropy E (the w of fused low-frequency image at this time1,w2) fitness function as Chaos particle swarm optimization algorithm, using mixed
Ignorant particle swarm optimization algorithm adaptively finds best fusion weight factor w1And w2;
(3) in order to retain the detailed information of image to be fused, to the high frequency direction subband in the step (1), using partial zones
Convergence strategy of the domain energy as characteristic quantity:
Wherein, CFI(s,d)(x, y) is the high-frequency sub-band coefficient in the direction fused s scale d,With
High-frequency sub-band of the synthesis intensity image I and polarization characteristic image PF on the direction s scale d is respectively indicated centered on (x, y)
Energy of local area, calculation formula is as follows:
In formula: M, N indicate the length and width of the local rectangular portions centered on (x, y);
(4) right to fused low frequency sub-band coefficient in step (2) using the uneven illumination in Retinex algorithm removal image
Fused high frequency direction sub-band coefficients filter out noise using Adaptive Thresholding in step (3);Wherein, Retinex algorithm is adopted
It is multi-Scale Retinex Algorithm, Retinex algorithm is handled:
R (x, y) is output low frequency subband, and F (x, y) is Gaussian filter function, and S (x, y) is input low frequency sub-band, and K is scale
Number, akIt is the weight of each scale, K=3, and take a1=a2=a3=1/3,
Adaptive Wavelet Thrinkage method: here, the threshold value of each high frequency direction subband, which is chosen, uses following formula:
In formula, to most thin scale, l=4 is taken, remaining subband takes l=3, the noise criteria difference σ of original imagenWith mediant estimation of robust
Estimated, it may be assumed that σn=median (abs (E))/0.6745, the high-frequency sub-band parameter variance on the direction s scale dIt utilizes
The Monte Carlo estimation technique is calculated;
(5) transformation of HVS mask is carried out to treated in step (4) low frequency sub-band coefficient and high frequency direction sub-band coefficients, obtained
FDCT contrast masking coefficient based on human-eye visual characteristic;
(6) the FDCT contrast masking coefficient that the step (5) obtains is automatically processed using nonlinear mapping function, is dashed forward
The minutia of image out;
(7) FDCT low frequency sub-band coefficient is automatically adjusted using nonlinear gain function, improves image overall contrast ratio;
(8) contravariant of HVS mask is carried out through nonlinear mapping function treated FDCT contrast masking coefficient in step (6)
It changes, the FDCT high frequency direction sub-band coefficients that obtain that treated;
(9) FDCT contravariant is carried out to treated in step (7) and (8) FDCT low frequency sub-band coefficient and high frequency direction sub-band coefficients
It changes, obtains final polarization fusion results image FI.
2. a kind of underwater Polarization Image Fusion system based on bionic visual mechanism according to claim 1, feature exist
In: the second level Multiscale Fusion method, it is characterised in that: the HVS mask transformation in the step (5), including brightness
Two step of mask and contrast masking, first carries out luminance mask, the FDCT luminance mask coefficient after obtaining luminance mask are as follows:
Then degree of comparing mask, the FDCT contrast masking coefficient obtained after contrast masking are as follows:
In above-mentioned luminance mask and contrast masking formula, x(1,1)For the low frequency sub-band coefficient of FDCT, x(s,d)For the s ruler of FDCT
High frequency direction sub-band coefficients on degree, the direction d, s are scale, and d is subband direction, as s > 1, indicate high-frequency sub-band.
3. a kind of underwater Polarization Image Fusion system based on bionic visual mechanism according to claim 1, feature exist
In the second level Multiscale Fusion method, the step (6), the FDCT that is obtained after nonlinear mapping function is handled
Contrast masking coefficient are as follows:
Wherein,CLCM(s,d)FDCT comparison before indicating nonlinear mapping function processing
Spend mask coefficient.
4. a kind of underwater Polarization Image Fusion system based on bionic visual mechanism according to claim 1, feature exist
In the second level Multiscale Fusion method, step (7) through nonlinear gain function treated FDCT low frequency sub-band
Coefficient are as follows:
Wherein,x(1,1)For the FDCT low frequency sub-band coefficient before nonlinear gain function processing.
5. a kind of underwater Polarization Image Fusion system based on bionic visual mechanism according to claim 2, feature exist
HVS mask inverse transformation in the second level Multiscale Fusion method, the step (8) include anti-contrast masking and
Anti- two step of luminance mask, first obtaining FDCT luminance mask coefficient through anti-contrast mask process is
Obtaining FDCT high frequency direction sub-band coefficients through anti-luminance mask again is
6. a kind of underwater Polarization Image Fusion system based on bionic visual mechanism according to claim 1, feature exist
In: the second level Multiscale Fusion method, it is characterised in that: in the step (1), FDCT is carried out to underwater polarization image
Transformation, taking the level of transformation is ceil (log2(min (High, Width)) -3) layer, wherein High, Width respectively indicate figure
The height and width of picture.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611047282.6A CN106504222B (en) | 2016-11-21 | 2016-11-21 | A kind of underwater Polarization Image Fusion system based on bionic visual mechanism |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611047282.6A CN106504222B (en) | 2016-11-21 | 2016-11-21 | A kind of underwater Polarization Image Fusion system based on bionic visual mechanism |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106504222A CN106504222A (en) | 2017-03-15 |
CN106504222B true CN106504222B (en) | 2019-09-06 |
Family
ID=58328465
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611047282.6A Active CN106504222B (en) | 2016-11-21 | 2016-11-21 | A kind of underwater Polarization Image Fusion system based on bionic visual mechanism |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106504222B (en) |
Families Citing this family (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107024275B (en) * | 2017-04-05 | 2018-04-06 | 长春理工大学 | Imitate mantis shrimp eye multichannel more spectral coverage polarization imaging detection devices in real time |
CN106989822B (en) * | 2017-04-05 | 2018-04-06 | 长春理工大学 | Imitate the optional polarization state detection device of mantis shrimp eye spectral coverage and method |
CN107292859B (en) * | 2017-05-27 | 2019-12-24 | 西安电子科技大学 | Chaotic medium polarization image acquisition method based on optical correlator |
IT201700067132A1 (en) * | 2017-06-19 | 2018-12-19 | Scuola Superiore Di Studi Univ E Perfezionamento Santanna | METHOD AND DEVICE FOR THE RECOVERY AND IMPROVEMENT OF PHOTOGRAPHIC IMAGES |
CN107909547B (en) * | 2017-09-30 | 2020-11-20 | 天津大学 | Polarization image denoising method based on Stokes polarization image correlation |
CN108171672B (en) * | 2018-01-10 | 2019-06-21 | 西北工业大学 | Underwater optics Intellisense method based on red channel and full convolutional neural networks |
WO2020029178A1 (en) * | 2018-08-09 | 2020-02-13 | 太平洋未来科技(深圳)有限公司 | Light and shadow rendering method and device for virtual object in panoramic video, and electronic apparatus |
CN109949233B (en) * | 2019-02-18 | 2022-12-13 | 深圳蓝影医学科技股份有限公司 | Method, system, device and storage medium for filtering scattered rays in X-ray image |
CN111369533B (en) * | 2020-03-05 | 2023-06-06 | 中国铁道科学研究院集团有限公司基础设施检测研究所 | Rail profile detection method and device based on polarization image fusion |
CN111369484B (en) * | 2020-03-05 | 2023-05-09 | 中国铁道科学研究院集团有限公司基础设施检测研究所 | Rail profile detection method and device |
CN112070698B (en) * | 2020-09-07 | 2022-08-05 | 合肥工业大学 | Target and background contrast enhancement method based on multi-channel polarization distance model |
CN112132763B (en) * | 2020-09-22 | 2023-10-20 | 中国科学院合肥物质科学研究院 | Underwater image restoration method and system based on human eye double antagonism color constancy |
CN112907472A (en) * | 2021-02-09 | 2021-06-04 | 大连海事大学 | Polarization underwater image optimization method based on scene depth information |
CN115200711B (en) * | 2022-06-02 | 2023-07-11 | 哈尔滨理工大学 | Monte Carlo algorithm for vector beam transmission in complex environment and polarization imaging method |
CN115631123B (en) * | 2022-11-22 | 2023-03-03 | 北京航空航天大学 | Bionic vision fusion severe environment imaging device and method |
CN116659520B (en) * | 2023-08-02 | 2023-10-20 | 中国人民解放军国防科技大学 | Matching positioning method, device and equipment based on bionic polarization vision enhancement |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101120359B (en) * | 2005-02-02 | 2011-06-15 | 因特格莱弗软件技术公司 | Real-time image detection using polarization data |
CN103900696B (en) * | 2014-03-07 | 2015-09-30 | 河海大学 | The method of polarization imaging under water of a kind of imitative squill vision polarization antagonism perception |
CN104504664B (en) * | 2014-12-29 | 2017-03-15 | 河海大学常州校区 | The automatic strengthening system of NSCT domains underwater picture based on human-eye visual characteristic and its method |
CN104657965B (en) * | 2015-03-12 | 2018-01-12 | 长春理工大学 | Polarization image fusion method based on discrete continuous Qu Bo |
-
2016
- 2016-11-21 CN CN201611047282.6A patent/CN106504222B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN106504222A (en) | 2017-03-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106504222B (en) | A kind of underwater Polarization Image Fusion system based on bionic visual mechanism | |
Bai et al. | Underwater image enhancement based on global and local equalization of histogram and dual-image multi-scale fusion | |
Schettini et al. | Underwater image processing: state of the art of restoration and image enhancement methods | |
Meng et al. | Efficient image dehazing with boundary constraint and contextual regularization | |
Lu et al. | Underwater image enhancement using guided trigonometric bilateral filter and fast automatic color correction | |
Rajkumar et al. | A comparative analysis on image quality assessment for real time satellite images | |
Yang et al. | Offshore underwater image restoration using reflection-decomposition-based transmission map estimation | |
CN108122206A (en) | A kind of low-light (level) image denoising method and device | |
CN106981053A (en) | A kind of underwater picture Enhancement Method based on Weighted Fusion | |
Reddy et al. | Image and signal processing in the underwater environment | |
CN109636766A (en) | Polarization differential and intensity image Multiscale Fusion method based on marginal information enhancing | |
Zhou et al. | A multifeature fusion method for the color distortion and low contrast of underwater images | |
Sun et al. | Underwater image enhancement with encoding-decoding deep CNN networks | |
Zhang et al. | Single image dehazing based on fast wavelet transform with weighted image fusion | |
CN103065291A (en) | Image fusion method based on promoting wavelet transform and correlation of pixel regions | |
Chang | Single underwater image restoration based on adaptive transmission fusion | |
Mathur et al. | Enhancement algorithm for high visibility of underwater images | |
Mathur et al. | Enhancement of nonuniformly illuminated underwater images | |
Qian et al. | Underwater image recovery method based on hyperspectral polarization imaging | |
Guo et al. | Haze removal for single image: A comprehensive review | |
Fu et al. | An anisotropic Gaussian filtering model for image de-hazing | |
CN109509201A (en) | A kind of SAR image quality evaluating method and device | |
Gong et al. | Research on the method of color compensation and underwater image restoration based on polarization characteristics | |
Yang et al. | Multi-scale underwater image enhancement with optimized homomorphic filter and RGB color correction | |
Natarajan | A review on underwater image enhancement techniques |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | 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 |