CN104050678A - Underwater monitoring color image quality measurement method - Google Patents
Underwater monitoring color image quality measurement method Download PDFInfo
- Publication number
- CN104050678A CN104050678A CN201410309507.5A CN201410309507A CN104050678A CN 104050678 A CN104050678 A CN 104050678A CN 201410309507 A CN201410309507 A CN 201410309507A CN 104050678 A CN104050678 A CN 104050678A
- Authority
- CN
- China
- Prior art keywords
- image
- underwater
- quality
- color image
- color
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Image Processing (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The invention discloses an underwater monitoring color image quality measurement method. The variance of hue and saturation of the CIELab space and the weighted array of brightness definition are utilized for measuring the quality of an underwater monitoring color image; the hue variance, the saturation variance and the brightness space definition of the CIELab space are calculated; then multiple linear regression MPL is applied to a plurality of underwater monitoring and investigation color images in a subjective evaluation test set to obtain three weighting coefficients, finally the quality evaluation UCIQE value of the underwater color image I is obtained, and the UCIQE value is directly proportional to the quality of the underwater color image. The method is reasonable, the quality difference of the content of a similar underwater color image can be described more accurately, and the method is more suitable for evaluating typical degradation of the underwater color image. The underwater monitoring color image quality measurement method is small in calculated amount and high in speed, and the real-time performance of the practical application of an algorithm is improved.
Description
Technical field
The invention belongs to image and process and analysis technical field, particularly processing and the analysis of color image shot in the environment that has absorption and diffuse optical decay.Relate to without measuring and the mass measuring method of coloured image under water with reference to color image quality.
Background technology
Along with the continuous progress of underwater sensor technical merit, how improving underwater video image information automatic processing capabilities is the key that solves intelligent underwater monitoring and underwater operation.A kind of method reliable and rational color image quality objective evaluation under water is applied for the processing of various underwater pictures, video, for example: in underwater picture classification, Color Image Analysis, Underwater Targets Recognition, underwater image restoration and enhancing etc. under water, automatic Prediction perception underwater picture quality is very important.In image quality evaluation research, although subjective assessment is considered to rationality and the highest method of accuracy, because its length consuming time, cost is high, process is loaded down with trivial details etc. can not well be applied in practice.Objectively image quality evaluating method can be divided into full reference type (Full-Reference, FR), partial reference type (Reduced-Reference, RR) and without three kinds of reference types (No-Reference, NR).Full reference type supposition original image exists and can obtain completely, and evaluation result is also more feasible.The Partial Feature of partial reference type hypothesis original image is known.But airborne image is different from water, in underwater environment, do not have original image to say, the underwater picture obtaining for in-site measurement, we need one without the objective image quality evaluation method with reference to blind evaluation in other words, are used for defining real picture quality in varying environment.Set up such standard very important, computer system need to know when stop, and determines whether had been found that best result, and the result of this " best " is acceptable in comparable underwater environment.Although the research to Color Image Processing is under water like a raging fire in recent years, there is no a kind of method for evaluating objective quality of coloured image under water that is applicable to.Typical coloured image is under water because different lighting conditions and water body make mostly to have following problem to absorption effect and the scattering effect etc. of light: limited visual range, low contrast, non-uniform lighting, fuzzy, hot spot, color projection and various Complex Noise.Because underwater picture is different from the degeneration of natural image, existing color image quality method for objectively evaluating cannot be effectively for evaluating the quality of coloured image under water.And degenerate in various degree and strengthen that effect is more difficult often afterwards for evaluating human eye discernible coloured image under water.
Majority designs mainly for natural gray image JPEG compression artefacts or specific application the research of reference-free quality evaluation method at present, comprises the method based on contrast and the method based on marginal sharpness.Due to the nonlinear characteristic of human visual system to different color, the visually-perceptible quality that makes to measure a secondary coloured image is very difficult.The expansion that existing most of natural color image evaluation method is gray level image evaluation method.After coloured image being converted to gray level image is evaluated or the image that coloured image is considered as three passages is applied respectively to grayscale image quality evaluation method, carry out weighted array.But the conversion from coloured image to gray level image itself is exactly the process of a distortion, so this evaluation method is obviously unaccommodated.For example also have some coloured image evaluation methods, for the evaluating in a certain respect of coloured image, entropy, brightness, sharpness, contrast and tone.For example Halser and susstunk attempt the variation of quantitatively evaluating natural color image at compression tone after treatment, by rg and Huang-blue space yb between R, G, B linear combination Cheng Hong-green sky, adopt the combination of associating variance and average as the method for evaluating coloured image tonal variation.On this basis, the people such as Fu adopt similar method, propose the combination of a kind of tone, sharpness and contrast as the method (Color Image Quality Index, CIQI) of evaluating color image quality.They adopt the statistical variance of rg and yb and average as tone, adopt the result of Sobel rim detection as sharpness, maximum local contrast is spent as a comparison, but this expression and subjective perception do not show linear relationship.The people such as Karen are on the basis of CIQI, weighted array color image quality evaluation method (the Color Quality Enhancement of another kind of tone, sharpness and contrast based on logarithm has been proposed, CQE), and by gray level image contrast (Root Mean Enhancement, RME) evaluate and expand to the multiple passages of coloured image, carry out modeling to the colored variance of the structure between color plane and color part is different with structure, a kind of colour contrast evaluation method (Color RME, CRME) has been proposed.
The reason that the process natural image degenerations such as the degeneration of coloured image under water and JPEG compression produce has a great difference.Not only there is the decline of caused by absorption and scattering fuzzy, contrast in coloured image under water, and have color projection degradation phenomena heterogeneous, and the absorption that this grows Different lightwave with water body is relevant with the spectral distribution of decay and target range and light source.Existing natural color image evaluation method cannot the effectively evaluating color projection problem of coloured image under water, and calculation of complex.
The degeneration factor more complicated of underwater picture, different from the quality assessment of natural image, there is not yet so far underwater picture database and can be for reference to subjective assessment value.For the little by little raising of quality obtaining in enhancing or recovery serious degradation image process, be difficult to judge that a kind of method is better than another kind of result, and for real-time and automatic processing, this is but most important.In the processing of gray level image, the people such as Schechner and N. Karpel think that contrast is a reasonably standard for tolerance underwater picture quality in polarized light device analysis under water under water.To taking back scattering as main, there is the submarine target image of noise in the people such as Hou, has proposed a kind of image definition evaluation standard based on weighting gray scale yardstick angle (GSA, GrayScale Angle).Arnold-Bos etc., in deep-sea and the experiment of unknown aquatic environment, have adopted the interpretational criteria of Pratt, think that, to better contrast and noise-free picture, gradient magnitude histogram should be similar to exponential distribution, have a small leak at low gradient place.Their robust index between having defined one 0 ~ 1.On the contrary, if image is subject to noise pollution serious, or contrast is very low, and gradient magnitude distribution is different.The people such as Arredondo have proposed the method for qualitative assessment underwater noise to underwater video image, the angular difference between the speed of estimating and true velocity is divided, for evaluating underwater video image.And in the enhancing of submerged color image and processing, to the evaluation of underwater picture quality, still taking subjective assessment as main, or the raising ratio that adopts SNR and MSE and subsequent characteristics to extract accuracy rate is weighed the validity of colour-image reinforcing method.At present, do not have a kind ofly for the color image quality evaluation method of coloured image degeneration factor under water, and can meet the real-time demand of underwater monitoring work.
Summary of the invention
The technical problem to be solved in the present invention be for habitata, submarine line detect and other Underwater Engineering investigation in the mass measurement problem of coloured image under water, propose one under water coloured image without reference mass measuring method.The method combines observer to the responsive statistical attribute of coloured image under water, the incompatible measurement habitata of set of weights, the submarine line of application CIELab color space color, saturation degree variance and the average sharpness of piece detect and other Underwater Engineering investigation in typical color image quality.The underwater monitoring image that the present invention is applicable to exist that saturation degree declines, the projection of non-homogeneous color, fuzzy and extra large snow are degenerated and other have the optical attenuation environment of identical degeneration.Computation complexity is low, on the different distortion level of the similar content under water of prediction, has higher accuracy, better with the average subjective achievement MOS degree of correlation, and real-time.
Technical matters to be solved by this invention is to realize by following technical scheme.The present invention is a kind of underwater monitoring color image quality measuring method, is characterized in, the method adopts the color harmony saturation degree variance in CIELab space and the weighted array of brightness sharpness to measure the quality of coloured image under water; Its step is as follows:
If
i p for CIELab space coloured image pixel under water,
p=1
n,
nfor the number of pixels in image;
i p =[
l p a p b p ],
l p ,
a p ,
b p be respectively CIELab spatial brightness and color component;
The first step, is calculated as follows CIELab space tone
c p , CIELab space saturation degree
s p ,
（1）
（2）
Second step, calculates the tone variance in CIELab space
var_C p with saturation degree variance
var_S p ,
（3）
（4）
（5）
（6）
The 3rd step, calculates brightness space sharpness
l_sharpness:
(7)
Wherein,
n edge for the amount of edge of brightness space Sobel rim detection, if sum is greater than threshold value
threshold 1 , think in image and have target, image is divided into
n×
nthe image block of size; If image block
p i edge pixel number in piece is greater than
threshold 2, be chosen to be the image block of evaluative meaning,
lfor there being the number of evaluative meaning image block; To significant image block, calculate
p i sharpness in piece:
. (8)
Wherein,
k 1 ×
k 2 for concrete block size,
f ij for pixel in piece (
i,
j) gray-scale value; Choosing of threshold value is relevant with the size of image block with the size of image;
threshold 1 ,
threshold 2 be less than 10% of image and block size;
The 4th step, after calculating tone, saturation degree variance and brightness space sharpness, obtains three weighting coefficients to the some width underwater monitorings in subjective assessment test set and investigation coloured image application multiple linear regression MPL
c 1 ,
c 2 ,
c 3 ,
c 1 +
c 2 +
c 3 =1.
Finally, obtain coloured image under water
iquality assessment value UCIQE be:
（9）
uCIQE[0,1],
uCIQEvalue is directly proportional to the quality of coloured image under water.
In a kind of underwater monitoring color image quality measuring method of the present invention: described
threshold 1 ,
threshold 2 preferably be less than 10% of image and block size.
In a kind of underwater monitoring color image quality measuring method of the present invention, image can be divided into
n×
nsize image block time, described
n=32 or 56 or 72.
The present invention be directed to simultaneously exist absorb, fuzzy, contrast that scattering causes reduce, saturation degree reduces and Underwater Engineering investigation, the monitoring coloured image of the degeneration factor such as non-homogeneous color projection.Principle of the present invention is: light in water the absorption in transmitting procedure and diffuse transmission influence the effect of whole Underwater Imaging.Forward scattering causes the fuzzy of characteristics of image, and back scattering makes the contrast of image reduce conventionally, produces vaporific fuzzy being superimposed upon on image.Although can increase visual range by increasing artificial light, often can cause non-uniform lighting situation, in image, produce speck, and speck is very dark around.Along with the increase of the degree of depth under water, color disappears successively according to wavelength, be approximately 100 times near blue end spectral absorption speed in the infiltration rate of the wavelength near the red one end of spectrum, approximately the place of 3m is red under water disappears, the local orange color dissipated of 5m, most of yellow in 10m dark left and right disappearance, last green and purple disappear in darker place, and blue because wavelength is the shortest, the distance of propagating is under water the longest, in 200m degree of depth left and right, all colors all can disappear.Except due to the decay of propagation distance under water, in water, the size of particulate and attribute affect the speed of scattering, reflection, propagation and absorption too.And in all impacts under water in the factor of color representation, absorption is most important.Therefore, the comprehensive degeneration of coloured image under water shows as saturation degree and reduces, non-homogeneous color projection, and contrast reduces, the fuzzy and noise of details.Accompanying drawing 1-6 is one group of underwater monitoring and investigates coloured image and corresponding polar coordinates hue histogram under water.Can find out the difference due to water body attenuation by absorption and light source, the tone skewness of coloured image under initial condition, the inclined to one side blue-green of tone of most underwater pictures.
During image is processed and analyzed under water, we need so a kind of quality evaluating method: (a) relevant to visual perception; (b) be reliable to underwater picture enhancing and underwater monitoring task; (c) for the typical degenerative character of coloured image under water; (d) can predict accurately the different distortion levels between similar content underwater picture; (e) there is low computation complexity and can be used for real-time system.
CIE L*a*b*(CIELab) be the uniform colour space that is usually used for describing the visible all colours of human eye, and be device-independent.Halser and susstunk have studied the distribution of CIELab space coloured image, and suppose that the color of image can describe with the linear combination of a series of CIELab spatial statistics information.Comprising: a, b component variance
,
, ab space variance hypotenuse length
, colourity average
and variance
, saturation degree average
and variance
deng 12 statistics components.By selecting different statistical informations to combine the color of Description Image.For the method for coloured image color distortion in research effective evaluation underwater monitoring, we have organized 12 observers, to existing, color projection, saturation degree decline, the comprehensive underwater monitoring of degenerating of fuzzy and extra large snow, engineering sounding image totally 44 width carried out subjective test and appraisal, test water hypograph comprises second-rate pipe inspection, coloury submarine photography, comprise natural lighting, laser, LED white light source, details is enriched and the simple polytype that waits.Partial test image is as shown in accompanying drawing 1-3.To each subjective assessment grade CIELab space colourity average of coloured image under water
and variance
, saturation degree average
and variance
change totally 5 statistical measures with luminance contrast and analyze, observers think preferably coloured image rich color under water as can be seen from the results, and luminance contrast, tone variance, saturation degree variance exceed average level.Change with subjective assessment, the ratio that these three components exceed average declines gradually.The distribution of test set image C IELab spatial statistics metric averaging values at different levels, can find out, the mean value of tone variance, saturation degree mean variance and luminance contrast is with subjective assessment grade linear growth.Observers are closely related on the visually-perceptible of coloured image and the impact of tone variance, saturation degree variance and contrast under water.
In addition, the attenuation by absorption of the single and water body of underwater picture background, causes coloured image tone distribution under water to concentrate on blue-green region, and tone distributes and concentrates.Therefore, in UCIQE, adopt tone variance to describe the COLOR COMPOSITION THROUGH DISTRIBUTION of coloured image under water.Meanwhile, in the time existing light source to irradiate in muddy water, the sea snow noise that water float particle causes is to affect another factor of color image quality under water.Adopt the common tolerance based on contrast and gradient can be subject to the impact of " Hai Xue " noise, cannot correctly evaluate.And tone in coloured image is portrayed the main distribution of color and can not be subject to the impact of particle in water, accompanying drawing 7 is in the transparency tank experiments that is 680cm, the target image sequence increasing with camera distance under artificial light source irradiates, tone in image distributes can describe with camera distance as seen from Figure 8 increases the linear image degeneration producing, Fig. 9 is that corresponding hue histogram distributes, can find out with camera distance and increase by being presented in Fig. 9 data, the variance of tone reduces gradually.
The brightness of underwater picture is lower, having in the coloured image of low-light level contrast, based on picking out details to the difference of color saturation from background.Accompanying drawing 10-13 is respectively in 202cm transparency tank experiments environment, the iron pipe of different cameral distance and target coloured image and corresponding saturation degree key map thereof under artificial light source irradiates, can find out, the highlighted reflection of coloured image under water being caused by illumination is corresponding with the maximum value region in saturation degree image.Therefore in the dark underwater environment having under light source irradiation, the non-uniform brightness phenomenon in the underwater picture that illumination causes can be reflected by saturation degree.
The present invention be coloured image under water based on CIELab spatial color, saturation degree variance and the even sharpness of piece without reference mass measuring method (Underwater Color Image Quality Evaluation, UCIQE), can be used in objective appraisal underwater monitoring task the color image quality of fuzzy, the non-homogeneous color projection that the absorption to different wavelengths of light and scattering cause by water body.Can be used for the Real-Time Evaluation colored effect that strengthens and restore under water, and can be used for design and realize the automatic adjustment of parameter in underwater picture disposal system and obtain the view data of best in quality.The present invention is for evaluating the color projection that exists light source and water body to cause, submarine line monitoring, the engineering sounding coloured image that fuzzy and extra large snow noise is degenerated.The present invention not only can be for the quality evaluation of underwater picture, being equally applicable to other exists in the image forming medium that optical scattering is relevant with decay, the image of for example, taking under greasy weather, smog environment, the present invention also can, for medical imaging, be subject to biological scattering medium as blood and organize the image affecting and take for evaluating.
By the foundation of underwater picture Environmental Evaluation Model, the operating personnel of marine systems can optimize the imaging system of different configurations, and post-processing approach different in Underwater Imaging is provided to suitable suggestion; From a large amount of video images, selecting automatically image can, with still abandoning, be very significant for the automatic classification and the identification target that meet real-time.Particularly, set up effective objective measurement method to underwater picture quality, to dynamically detecting and adjust picture quality, automatically adjust parameter in image capturing system and obtain the view data of best in quality.According to the result of mass measurement, image being carried out to image restoration, image reconstruction, rear filtering scheduling algorithm carries out optimal treatment and has important value as the tolerance of the algorithm quality of judging image processing system.
The present invention compared with prior art has the following advantages:
(1) the present invention proposes the first method that color image quality is measured that can be used under water.
(2) the present invention is mainly for the saturation degree decline in Underwater Engineering investigation, surveillance map picture, the projection of non-homogeneous color, the comprehensive degradation phenomena of fuzzy and extra large snow.
(3) the present invention, compared with existing main color image quality evaluation method, can describe the difference in quality of similar color image content under water more accurately, is more suitable for and evaluates typical case's degeneration of coloured image under water.
(4) the present invention adopts block margin judgment mechanism, can weaken the impact of noise on the judgement of target image quality in underwater picture, has certain robustness.
(5) the present invention carries out block parallel processing to target image, and brightness and color to CIELab space calculate respectively, has improved the real-time of algorithm practical application.
(6) calculated amount of the present invention is little, and speed is fast.
(7) in the present invention, adopt the mass measuring method of weighted array, when in conjunction with underwater video product, can design and regulate selection, met the needs of different underwater environments.
Brief description of the drawings
Fig. 1-3 are underwater picture exemplary plot;
Fig. 4-6 are Fig. 1-3 polar coordinates hue histograms of correspondence successively;
Fig. 7 is artificial light source pond target image sets;
Fig. 8 is the tone illustration group that Fig. 7 is corresponding;
Fig. 9 is the hue histogram distribution plan group that Fig. 7 is corresponding;
Figure 10 is artificial light source pond iron pipe experimental image group;
Figure 11 is the saturation degree key map group of Figure 10;
Figure 12 is artificial light source pond target experimental image group;
Figure 13 is Figure 12 and saturation degree key map group;
Figure 14 is pond internal object figure;
Figure 15 is pond internal object image sets in natural lighting 680cm transparency Figure 14;
Figure 16 is the image quality evaluation result figure that Figure 15 adopts CRME;
Figure 17 is the image quality evaluation result figure that Figure 15 adopts CQE;
Figure 18 is the image quality evaluation result figure that Figure 15 adopts CIQI;
Figure 19 is the image quality evaluation result figure that Figure 15 adopts UCIQE;
Figure 20 is another pond internal object figure;
Figure 21 is pond internal object figure group in natural light 365cm transparency Figure 20;
Figure 22 is the image quality evaluation result figure that Figure 21 adopts CRME;
Figure 23 is the image quality evaluation result figure that Figure 21 adopts CQE;
Figure 24 is the image quality evaluation result figure that Figure 21 adopts CIQI;
Figure 25 is the image quality evaluation result figure that Figure 21 adopts UCIQE.
Embodiment
Below further technical scheme of the present invention is described, makes those skilled in the art further understand the present invention, and do not form the restriction to right of the present invention.
Embodiment 1, a kind of underwater monitoring color image quality measuring method, the method adopts the color harmony saturation degree variance in CIELab space and the weighted array of brightness sharpness to measure the quality of coloured image under water; Its step is as follows:
If
i p for CIELab space coloured image pixel under water,
p=1
n,
nfor the number of pixels in image;
i p =[
l p a p b p ],
l p ,
a p ,
b p be respectively CIELab spatial brightness and color component;
The first step, is calculated as follows CIELab space tone
c p , CIELab space saturation degree
s p ,
（1）
（2）
Second step, calculates the tone variance in CIELab space
var_C p with saturation degree variance
var_S p ,
（3）
（4）
（5）
（6）
The 3rd step, calculates brightness space sharpness
l_sharpness:
(7)
Wherein,
n edge for the amount of edge of brightness space Sobel rim detection, if sum is greater than threshold value
threshold 1 , think in image and have target, image is divided into
n×
nthe image block of size; If image block
p i edge pixel number in piece is greater than
threshold 2, be chosen to be the image block of evaluative meaning,
lfor there being the number of evaluative meaning image block; To significant image block, calculate
p i sharpness in piece:
. (8)
Wherein,
k 1 ×
k 2 for concrete block size,
f ij for pixel in piece (
i,
j) gray-scale value; Choosing of threshold value is relevant with the size of image block with the size of image;
threshold 1 ,
threshold 2 be less than 10% of image and block size;
The 4th step, after calculating tone, saturation degree variance and brightness space sharpness, obtains three weighting coefficients to the some width underwater monitorings in subjective assessment test set and investigation coloured image application multiple linear regression MPL
c 1 ,
c 2 ,
c 3 ,
c 1 +
c 2 +
c 3 =1.
Finally, obtain coloured image under water
iquality assessment value UCIQE be:
（9）
uCIQE[0,1],
uCIQEvalue is directly proportional to the quality of coloured image under water.
Embodiment 2, described a kind of underwater monitoring color image quality measuring method, described
threshold 1 ,
threshold 2 be less than 10% of image and block size.
Embodiment 3, described a kind of underwater monitoring color image quality measuring method, is characterized in that, image is divided into
n×
nsize image block time, described
n=32 or 56 or 72.
Embodiment 4, a kind of underwater monitoring color image quality measuring method (Underwater Color Image Quality Evaluation, UCIQE).For have that fuzzy, the contrast that scattering causes reduces, saturation degree reduces and Underwater Engineering investigation, the monitoring coloured image of the degeneration factor such as non-homogeneous color projection simultaneously.Concrete grammar step is as follows:
If
i p for CIELab space coloured image pixel under water,
p=1
n,
nfor the number of pixels in image.
i p =[
l p a p b p ],
l p ,
a p ,
b p be respectively CIELab spatial brightness and color component,
c p for CIELab space tone,
s p for CIELab space saturation degree.Coloured image under water
iquality assessment value (Underwater Color Image Quality Evaluation, UCIQE) be,
（1）
Wherein,
uCIQE[0,1],
var_C p ,
var_S p for the color harmony saturation degree variance in CIELab space,
l_sharpnessfor brightness space sharpness,
c 1 ,
c 2 ,
c 3 weighting coefficient,
c 1 +
c 2 +
c 3 =1.
uCIQEbe worth greatlyr, illustrate that the quality of coloured image is better under water.
（2）
（3）
（4）
（5）
（6）
（7）
(8)
Wherein,
n edge for the amount of edge of brightness space Sobel rim detection, if sum is greater than threshold value
threshold 1 , think in image and have target, image is divided into
n×
nthe image block of size.(selection of block size can be 32 × 32,56 × 56,72 × 72 etc.), N=72 in the present embodiment.If image block
p i edge pixel number in piece is greater than
threshold 2, be chosen to be the image block of evaluative meaning,
lfor there being the number of evaluative meaning image block.To significant image block, calculate
p i sharpness in piece:
. (9)
Wherein,
k 1 ×
k 2 for concrete block size,
f ij for pixel in piece (
i,
j) gray-scale value.Choosing of threshold value is relevant with the size of image block with the size of image.Experiment showed,
threshold 1 , and
threshold 2 be less than 10% of image and block size and can obtain good evaluation result, in this enforcement, get
threshold 1 =3%,
threshold 2 =10%.
After calculating tone, saturation degree variance and brightness space sharpness, 31 width underwater monitorings and investigation coloured image application multiple linear regression (MPL) in subjective assessment test set are obtained to three coefficients in (1) formula.In the present invention, to the color projection being caused by light source and water body, fuzzy, the muddy type submarine line monitoring of suspended particle, engineering sounding underwater picture,
c 1 =0.5162,
c 2 =0.1882,
c 3 =0.2956.Because calculating and tone, the saturation degree of sharpness are independently calculated, so three components can parallel processing carry out raising speed.To the coloured image under water of other types, if training image has similar degeneration factor, can carry out weight coefficient matching by corresponding multiple linear regression, make UCIQE obtain good performance.
Embodiment 5, the accuracy performance experiment that the coloured image under water that color image quality measuring method UCIQE degenerates in various degree to increasing with camera distance is under water evaluated:
In the image embodiment of pond, long 4.5 meters of pond, wide 2 meters, high 2 meters.Approximately 15 tons of each test water.Four observation windows have been worn in pond three.Test objective is iron pipe and target, and wherein target is the blank of 40 cm × 50 cm, and fringe spacing is above 2mm, and width of fringe is followed successively by 1mm, 2mm, 3 mm, 4 mm, in air as shown in Figure 14.Adopt OTI-UWC-325/P/E color camera, respectively at 680cm, under 365cm and 202cm transparency (Duntley rule) condition, take underwater picture sequence.
Accompanying drawing 15-25 be natural lighting increase gradually with camera distance in 680cm transparency, target image sequence and 365cm transparency iron pipe image sequence adopt the contrast of UCIQE of the present invention and other three kinds of Major Natural coloured image evaluation methods under water, from the variation of four kinds of method evaluation results, with the continuous increase of camera distance, optical attenuation degree is increasing, and UCIQE of the present invention linear reflection more accurately changes the decline of underwater picture quality with camera distance.
Embodiment 6, the degree of correlation of color image quality measuring method UCIQE and subjective judgement experiment under water:
Subjective testing process is as follows:
In the set of subjective assessment test pattern, submarine line detects, Underwater Engineering investigation class image totally 44 width, comprehensively has the multiple degradation phenomenas such as color projection, saturation degree decline, fuzzy and Hai Xue.Observer is not the professional of image and field of video processing, and does not participate in similarly subjective assessment.Under test water, coloured image is according to random played in order, and for reducing the subjectivity rate that peels off, every width image is random to be repeated 4 times.Observer adopts Pyatyi quality score list advocate approach processed to carry out quality assessment to test coloured image.Quality grade scoring is: non-constant (1), poor (2), general (3), good (4) and very good (5).In order to obtain significant analysis, in experimental result comparison process, we carry out correlation computations with different color image datas under water.We adopt 31 width image optimization weighting coefficients, relevant by 13 remaining width image calculation.Adopt 4 parameter 3 polynomial expressions to carry out regretional analysis, calculate the correlativity of prediction subjective quality assessment value and average subjective scores MOS.Employing Pearson linearly dependent coefficient (Pearson ' s product moment correlation, PRCC) departure degree of measurement objective evaluation value and MOS, root-mean-square error (Root mean square error, RMSE) contrast the accuracy of the UCIQE quality assessment of proposition, Spearman coefficient of rank correlation (Spearman ' s rank order correlation, ROCC) is measured the monotonicity of evaluation method
[33], comparing result is as shown in subordinate list 1.Can be found out by data in table 1, the UCIQE in the present invention exists muddy type submarine line monitoring, engineering sounding image and the subjective assessment (MOS) that color projects, fuzzy and suspended particle causes to have higher correlativity for evaluating, and reaches 0.84.Exceed 63%, CQE than CIQI and exceed 69%, CRME and exceed 6%, accuracy rate is higher, and monotonicity is better, compares with existing other natural color image quality evaluating methods, is more suitable for and evaluates this type of coloured image under water.
Table 1 UCIQE performance comparison
| PRCC | RMSE | SRCC | |
| UCIQE | 0.8407 | 0.0704 | 0.7418 |
| CQE | 0.2598 | 0.1255 | 0.3297 |
| CRME | 0.7848 | 0.0805 | 0.3187 |
| CIQI | 0.3074 | 0.1237 | 0.3901 |
4 parameter 3 polynomial expressions in the present embodiment:
（10）
Pearson linearly dependent coefficient in the present embodiment (Pearson ' s product moment correlation, PRCC):
（11）
Root-mean-square error (Root mean square error, RMSE) in the present embodiment:
(12)
Spearman coefficient of rank correlation in the present embodiment (Spearman ' s rank order correlation, ROCC)
(13)
Wherein, OS is objective score,
x i ,
y i represent respectively
ithe prediction subjective scores SS of individual cycle tests
pwith corresponding subjective scores SS, n represents total number of cycle tests,
,
represent SS
paverage with SS.
rX i ,
rY i after representing that the subjective and subjective achievement of prediction sorts by same sequence (ascending) respectively, the
ithe sequence number of individual achievement in sequence separately.
Embodiment 7, color image quality is measured UCIQE computing velocity contrast experiment under water:
Subordinate list 2 is the average operating time contrast of 60 width underwater pictures, and the size of image is 545 × 749 × 3, and allocation of computer is Intel i7 2.8Ghz dual core processor, 8G internal memory, and application matlab2012b carries out algorithm simulating.Can find out from simulation result, UCIQE arithmetic speed is the fastest, faster 5 times than the CQE method of same measurement sharpness, tone, contrast.The real-time application of more satisfied color video frequency image under water.
The contrast of table 2 average operating time
| UCIQE | CQE | CRME | CIQI | |
| Average operating time (s) | 0.2855 | 1.925 | 0.966 | 1.056 |
Claims (3)
1. a underwater monitoring color image quality measuring method, is characterized in that, the method adopts the color harmony saturation degree variance in CIELab space and the weighted array of brightness sharpness to measure the quality of coloured image under water; Its step is as follows:
If
i p for CIELab space coloured image pixel under water,
p=1
n,
nfor the number of pixels in image;
i p =[
l p a p b p ],
l p ,
a p ,
b p be respectively CIELab spatial brightness and color component;
The first step, is calculated as follows CIELab space tone
c p , CIELab space saturation degree
s p ,
（1）
（2）
Second step, calculates the tone variance in CIELab space
var_C p with saturation degree variance
var_S p ,
（3）
（4）
（5）
（6）
The 3rd step, calculates brightness space sharpness
l_sharpness:
(7)
Wherein,
n edge for the amount of edge of brightness space Sobel rim detection, if sum is greater than threshold value
threshold 1 , think in image and have target, image is divided into
n×
nthe image block of size; If image block
p i edge pixel number in piece is greater than
threshold 2, be chosen to be the image block of evaluative meaning,
lfor there being the number of evaluative meaning image block; To significant image block, calculate
p i sharpness in piece:
. (8)
Wherein,
k 1 ×
k 2 for concrete block size,
f ij for pixel in piece (
i,
j) gray-scale value; Choosing of threshold value is relevant with the size of image block with the size of image;
threshold 1 ,
threshold 2 be less than 10% of image and block size;
The 4th step, after calculating tone, saturation degree variance and brightness space sharpness, obtains three weighting coefficients to the some width underwater monitorings in subjective assessment test set and investigation coloured image application multiple linear regression MPL
c 1 ,
c 2 ,
c 3 ,
c 1 +
c 2 +
c 3 =1;
Finally, obtain coloured image under water
iquality assessment value UCIQE be:
（9）
uCIQE[0,1],
uCIQEvalue is directly proportional to the quality of coloured image under water.
2. a kind of underwater monitoring color image quality measuring method according to claim 1, is characterized in that: described
threshold 1 ,
threshold 2 preferably be less than 10% of image and block size.
3. a kind of underwater monitoring color image quality measuring method according to claim 1 and 2, is characterized in that: image is divided into
n×
nsize image block time, described
n=32 or 56 or 72.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410309507.5A CN104050678A (en) | 2014-07-01 | 2014-07-01 | Underwater monitoring color image quality measurement method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410309507.5A CN104050678A (en) | 2014-07-01 | 2014-07-01 | Underwater monitoring color image quality measurement method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104050678A true CN104050678A (en) | 2014-09-17 |
Family
ID=51503461
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410309507.5A Pending CN104050678A (en) | 2014-07-01 | 2014-07-01 | Underwater monitoring color image quality measurement method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104050678A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108572008A (en) * | 2018-04-21 | 2018-09-25 | 张冬梅 | Tap lower reservoir electronics cleaning systems |
| CN108985350A (en) * | 2018-06-27 | 2018-12-11 | 北京中安未来科技有限公司 | It is a kind of that the method and apparatus of blurred picture are identified based on gradient magnitude sparse features information, calculate equipment and storage medium |
| CN109961073A (en) * | 2019-03-29 | 2019-07-02 | 云南电网有限责任公司电力科学研究院 | The acquisition methods and device of a kind of transmission line of electricity and shaft tower information |
| CN110378893A (en) * | 2019-07-24 | 2019-10-25 | 北京市博汇科技股份有限公司 | Image quality evaluating method, device and electronic equipment |
| CN111505659A (en) * | 2020-05-08 | 2020-08-07 | 中国科学院半导体研究所 | Underwater water body noise calculation three-dimensional imaging method and imaging system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102663745A (en) * | 2012-03-23 | 2012-09-12 | 北京理工大学 | Color fusion image quality evaluation method based on vision task. |
| CN103617617A (en) * | 2013-12-05 | 2014-03-05 | 淮海工学院 | Underwater image quality evaluating and measuring method based on power spectrum description |
| CN103632369A (en) * | 2013-12-05 | 2014-03-12 | 淮海工学院 | Method for universally detecting quality of non-reference underwater images on basis of combination of block average definition |
-
2014
- 2014-07-01 CN CN201410309507.5A patent/CN104050678A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102663745A (en) * | 2012-03-23 | 2012-09-12 | 北京理工大学 | Color fusion image quality evaluation method based on vision task. |
| CN103617617A (en) * | 2013-12-05 | 2014-03-05 | 淮海工学院 | Underwater image quality evaluating and measuring method based on power spectrum description |
| CN103632369A (en) * | 2013-12-05 | 2014-03-12 | 淮海工学院 | Method for universally detecting quality of non-reference underwater images on basis of combination of block average definition |
Non-Patent Citations (3)
| Title |
|---|
| ZHOU WANG 等: "A Universal Image Quality Index", 《IEEE SIGNAL PROCESSING LETTERS》 * |
| 杨杨: "基于均匀计色制的彩色图像质量评价", 《中国优秀硕士学位论文全文数据库 信息科技辑（月刊）》 * |
| 杨淼 等: "无参考水下静态图像质量评价与仿真分析", 《系统仿真学报》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108572008A (en) * | 2018-04-21 | 2018-09-25 | 张冬梅 | Tap lower reservoir electronics cleaning systems |
| CN108985350A (en) * | 2018-06-27 | 2018-12-11 | 北京中安未来科技有限公司 | It is a kind of that the method and apparatus of blurred picture are identified based on gradient magnitude sparse features information, calculate equipment and storage medium |
| CN109961073A (en) * | 2019-03-29 | 2019-07-02 | 云南电网有限责任公司电力科学研究院 | The acquisition methods and device of a kind of transmission line of electricity and shaft tower information |
| CN110378893A (en) * | 2019-07-24 | 2019-10-25 | 北京市博汇科技股份有限公司 | Image quality evaluating method, device and electronic equipment |
| CN110378893B (en) * | 2019-07-24 | 2021-11-16 | 北京市博汇科技股份有限公司 | Image quality evaluation method and device and electronic equipment |
| CN111505659A (en) * | 2020-05-08 | 2020-08-07 | 中国科学院半导体研究所 | Underwater water body noise calculation three-dimensional imaging method and imaging system |
| CN111505659B (en) * | 2020-05-08 | 2023-10-03 | 中国科学院半导体研究所 | Underwater dehydration body noise calculation three-dimensional imaging method and imaging system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106548467B (en) | The method and device of infrared image and visual image fusion | |
| Yang et al. | An underwater color image quality evaluation metric | |
| CN107172418B (en) | A kind of tone scale map image quality evaluating method based on exposure status analysis | |
| CN104050678A (en) | Underwater monitoring color image quality measurement method | |
| CN106447646A (en) | Quality blind evaluation method for unmanned aerial vehicle image | |
| CN105741328A (en) | Shot image quality evaluation method based on visual perception | |
| CN103974053B (en) | A kind of Automatic white balance antidote extracted based on ash point | |
| CN103617617B (en) | The underwater picture quality evaluation measuring method of describing based on power spectrum | |
| CN103632369B (en) | The combination of block average definition is without reference underwater picture universal property quantity measuring method | |
| CN106483130B (en) | A kind of detection method and its automatic detection device of rice disease | |
| CN106408526B (en) | A kind of visibility detecting method based on multilayer polar plot | |
| CN104198325B (en) | Stem ratio measuring method in pipe tobacco based on computer vision | |
| CN104680195B (en) | A kind of method of automatic identification road gate video and the vehicle color in picture | |
| CN109741285A (en) | A kind of construction method and system of underwater picture data set | |
| CN102565044B (en) | Method for detecting pesticide residue by adopting rapid-determination card on basis of image processing | |
| CN103076334A (en) | Method for quantitatively evaluating perceived quality of digital printed lines and texts | |
| CN106709903A (en) | PM2.5 concentration prediction method based on image quality | |
| CN105116412B (en) | A kind of wideband radar ISAR image properties quantitative identification method | |
| Luo et al. | An underwater-imaging-model-inspired no-reference quality metric for images in multi-colored environments | |
| CN110070480A (en) | A kind of analogy method of underwater optics image | |
| CN103200349A (en) | Scanned image color cast automatic detection method | |
| CN116402813B (en) | Neural network-based copper converter converting copper-making period end point judging method | |
| CN109035225A (en) | A kind of automotive brake pads presentation quality inspection illumination system layout quality evaluating method | |
| Li et al. | Underwater image sharpness assessment based on selective attenuation of color in the water | |
| CN112686105B (en) | Fog concentration grade identification method based on video image multi-feature fusion |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20140917 |