CN110276744A - The assessment method and device of image mosaic quality - Google Patents
The assessment method and device of image mosaic quality Download PDFInfo
- Publication number
- CN110276744A CN110276744A CN201910402871.9A CN201910402871A CN110276744A CN 110276744 A CN110276744 A CN 110276744A CN 201910402871 A CN201910402871 A CN 201910402871A CN 110276744 A CN110276744 A CN 110276744A
- Authority
- CN
- China
- Prior art keywords
- evaluated
- image
- attribute information
- information
- altimetric image
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformation in the plane of the image
- G06T3/40—Scaling the whole image or part thereof
- G06T3/4038—Scaling the whole image or part thereof for image mosaicing, i.e. plane images composed of plane sub-images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
- G06T7/0006—Industrial image inspection using a design-rule based approach
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30168—Image quality inspection
Abstract
The embodiment of the present invention provides the assessment method and device of a kind of image mosaic quality, the described method includes: obtaining the first attribute information of altimetric image to be evaluated and the second attribute information of reference picture, first attribute information includes the dead zone information of altimetric image to be evaluated, pixel information and frequency information, second attribute information includes the pixel information and frequency information of reference picture, altimetric image to be evaluated splices at least two first images and obtains, reference picture splices at least two second images and obtains, according to the first attribute information and second attribute information, determine the joining quality of altimetric image to be evaluated.For improving the accuracy tested and assessed to the joining quality of image to be evaluated.
Description
Technical field
The present embodiments relate to field of image processing more particularly to a kind of assessment methods and dress of image mosaic quality
It sets.
Background technique
Video camera can carry out Image Acquisition to the environment where it, and acquired image is spliced into panoramic picture,
So that user has impression on the spot in person when watching projection device broadcasting panoramic picture.
In practical applications, it after being spliced into panoramic picture, needs to test and assess to the joining quality of panoramic picture.Mesh
Before, the method to the assessment method of the joining quality of panoramic picture includes: to obtain reference picture and wait right on panoramic picture of testing and assessing
The gray value for answering the pixel of position, according to reference picture and wait the gray scale of the pixel of corresponding position on panoramic picture of testing and assessing
Value calculates the square mean error amount of reference picture and panoramic picture to be tested and assessed, and using the square mean error amount as panoramic picture
Joining quality.
In above process, square mean error amount is to be waited for according to the sum of the grayscale values of pixel on the reference picture on corresponding position
The gray value of pixel obtains on assessment panoramic picture, if on reference picture pixel with wait pixel on panoramic picture of testing and assessing
Position not to it is corresponding when lead to the square mean error amount obtained inaccuracy, and then lead to the assessment to the joining quality of panoramic picture
Inaccuracy.
Summary of the invention
The embodiment of the present invention provides the assessment method and device of a kind of image mosaic quality, for improving to image to be evaluated
The accuracy tested and assessed of joining quality.
In a first aspect, the embodiment of the present invention provides a kind of assessment method of image mosaic quality, comprising:
Obtain the first attribute information of altimetric image to be evaluated and the second attribute information of reference picture, first attribute information
Dead zone information, pixel information and frequency information including the altimetric image to be evaluated, second attribute information includes the ginseng
The pixel information and frequency information of image are examined, the altimetric image to be evaluated splices at least two first images and obtains, described
Reference picture splices at least two second images and obtains；
According to first attribute information and second attribute information, the joining quality of the altimetric image to be evaluated is determined.
It is described according to first attribute information and second attribute information in a kind of possible embodiment, really
The joining quality of the fixed altimetric image to be evaluated, comprising:
According to first attribute information and second attribute information, the feature vector of the altimetric image to be evaluated is determined；
Described eigenvector is handled, the joining quality of the altimetric image to be evaluated is obtained.
It is described according to first attribute information and second attribute information in alternatively possible embodiment,
Determine the feature vector of the altimetric image to be evaluated, comprising:
According to the dead zone information in first attribute information, the blind area feature of the altimetric image to be evaluated is determined；
According to the pixel information of pixel information and the second attribute information in first attribute information, determine described in
The histogram statistical features of altimetric image to be evaluated and sparse reconstruction features；
According to the frequency information of frequency information and the second attribute information in first attribute information, determine described to be evaluated
The perceived hash characteristics of altimetric image；
According to the altimetric image to be evaluated and at least two first image, determine that the color difference of the altimetric image to be evaluated is special
Sign；
According to the blind area feature, histogram statistical features, sparse reconstruction features, perceived hash characteristics and color difference feature,
Determine described eigenvector.
It is described that described eigenvector is handled in alternatively possible embodiment, obtain the mapping to be evaluated
The joining quality of picture, comprising:
Preset model is obtained, the preset model learns to obtain according to multiple groups sample image, every group of sampled images this packet
Include the feature vector and the corresponding sample joining quality of the sample image of sample image；
By the preset model, described eigenvector is handled, obtains the joining quality of the altimetric image to be evaluated.
In alternatively possible embodiment, the preset model are as follows:
Wherein,For the output valve of the preset model, β is the parameter matrix of the preset model, and x is the default mould
The feature vector to be entered of type.
Second aspect, the embodiment of the present invention provide a kind of assessment device of image mosaic quality, comprising: obtain module and really
Cover half block, wherein
The acquisition module is used for, and obtains the first attribute information of altimetric image to be evaluated and the second attribute letter of reference picture
Breath, first attribute information include dead zone information, pixel information and the frequency information of the altimetric image to be evaluated, and described second
Attribute information includes the pixel information and frequency information of the reference picture, and the altimetric image to be evaluated is splicing at least two the
What one image obtained, the reference picture splices at least two second images and obtains；
The determining module is used for, and according to first attribute information and second attribute information, is determined described to be evaluated
The joining quality of altimetric image.
In a kind of possible embodiment, the determining module is specifically used for:
According to first attribute information and second attribute information, the feature vector of the altimetric image to be evaluated is determined；
Described eigenvector is handled, the joining quality of the altimetric image to be evaluated is obtained.
In alternatively possible embodiment, the determining module is specifically used for:
According to the dead zone information in first attribute information, the blind area feature of the altimetric image to be evaluated is determined；
According to the pixel information of the altimetric image to be evaluated and the second attribute information, the histogram of the altimetric image to be evaluated is determined
Figure statistical nature and sparse reconstruction features；
According to the frequency information of frequency information and the second attribute information in first attribute information, determine described to be evaluated
The perceived hash characteristics of altimetric image；
According to the altimetric image to be evaluated and at least two first image, determine that the color difference of the altimetric image to be evaluated is special
Sign；
According to the blind area feature, histogram statistical features, sparse reconstruction features, perceived hash characteristics and color difference feature,
Determine described eigenvector.
In alternatively possible embodiment, the determining module is specifically used for:
Preset model is obtained, the preset model learns to obtain according to multiple groups sample image, every group of sampled images this packet
Include the feature vector and the corresponding sample joining quality of the sample image of sample image；
By the preset model, described eigenvector is handled, obtains the joining quality of the altimetric image to be evaluated.
In alternatively possible embodiment, the preset model are as follows:
Wherein,For the output valve of the preset model, β is the parameter matrix of the preset model, and x is the default mould
The feature vector to be entered of type.
The third aspect, the embodiment of the present invention provide a kind of assessment device of image mosaic quality, comprising: processor, it is described
Processor is coupled with memory；
The memory is used for, and stores computer program；
The processor is used for, and executes the computer program stored in the memory, so that described image splices matter
The assessment device of amount executes the assessment method of the described in any item image mosaic quality of above-mentioned first aspect.
Fourth aspect, the embodiment of the present invention provide a kind of readable storage medium storing program for executing, including program or instruction, when described program or
When instruction is run on computers, the assessment method of the image mosaic quality as described in above-mentioned first aspect any one is held
Row.
In the assessment method and device of image mosaic quality provided by the invention, the method obtains altimetric image to be evaluated
Second attribute information of the first attribute information and reference picture, the first attribute information include altimetric image to be evaluated dead zone information, as
Vegetarian refreshments information and frequency information, the second attribute information include the pixel information and frequency information of reference picture, altimetric image to be evaluated
At least two first images of splicing obtain, and reference picture splices at least two second images and obtains, and belong to according to first
Property information and the second attribute information, determine the joining quality of altimetric image to be evaluated.In above process, according to the first attribute information and
Second attribute information determines the joining quality of altimetric image to be evaluated, improves and tests and assesses to the joining quality of image to be evaluated
Accuracy.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is this hair
Bright some embodiments for those of ordinary skill in the art without any creative labor, can be with
It obtains other drawings based on these drawings.
Fig. 1 is the application scenario diagram of the assessment method of image mosaic quality provided in an embodiment of the present invention；
Fig. 2 is the flow chart one of the assessment method of image mosaic quality provided in an embodiment of the present invention；
Fig. 3 is the flow diagram two of the assessment method of image mosaic quality provided in an embodiment of the present invention；
Fig. 4 is the splicing of altimetric image to be evaluated and reference picture provided in an embodiment of the present invention；
Fig. 5 is the structural schematic diagram of the assessment device of image mosaic quality provided in an embodiment of the present invention.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art
Every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.
Fig. 1 is the application scenario diagram of the assessment method of image mosaic quality provided in an embodiment of the present invention.Referring to Figure 1,
For example, camera 10 includes camera 11 and camera 12, wherein the position of camera 11 and the positional symmetry of camera 12, phase
The image of the available ambient enviroment of machine 10, and according to the image got, determines panoramic picture, and the panoramic picture is to obtaining
The image got carries out what splicing obtained.
For example, when camera 10 is located at coordinate center O point, 11 face of camera, 0 degree of direction (12 face 180 degree side of camera
To) when, available first image 101 of camera 11, available first image 102 of camera 12, camera 10 can basis
First image 101 and the first image 102, determine panoramic picture 13.
For example, when camera 10 is located at coordinate center O point, 11 face of camera, 90 degree of directions (12 face of camera, 270 degree of sides
To) when, camera 11 can collect the second image 103, and camera 12 can collect the first image 104, camera 10
Panoramic picture 14 can be determined according to the first image 101 and the first image 102.
In practical applications, it usually needs the joining quality for the panoramic picture that camera 10 is determined is evaluated and tested.For example,
Panoramic picture 13 is altimetric image to be evaluated, then panoramic picture 14 can be reference picture, is carried out to the joining quality of panoramic picture 13
The method of evaluation and test includes: to determine altimetric image to be evaluated according to the attribute information of altimetric image 13 attribute information and reference picture 14 to be evaluated
13 blind area feature, histogram statistical features, sparse reconstruction features etc., and according to blind area feature, histogram statistical features, sparse
Reconstruction features etc. determine the joining quality of altimetric image 13 to be evaluated.
In above process, according to the blind area feature of altimetric image 13 to be evaluated, histogram statistical features, sparse reconstruction features
Deng determining the joining quality of altimetric image 13 to be evaluated, the accuracy of the assessment to the joining quality of panoramic picture can be improved.
In the following, technical solution shown in the application is described in detail by specific embodiment.Under it should be noted that
The several specific embodiments in face can be combined with each other, and for the same or similar content, no longer carry out weight in various embodiments
Multiple explanation.
Fig. 2 is the flow chart one of the assessment method of image mosaic quality provided in an embodiment of the present invention.Fig. 2 is referred to, is schemed
As the assessment method of joining quality includes:
S201: the first attribute information of altimetric image to be evaluated and the second attribute information of reference picture, the first attribute letter are obtained
Breath includes dead zone information, pixel information and the frequency information of the altimetric image to be evaluated, and the second attribute information includes reference picture
Pixel information and frequency information, altimetric image to be evaluated splice at least two first images and obtain, reference picture is splicing
What at least two second images obtained.
Optionally, the executing subject of the embodiment of the present invention can be with camera, or panoramic picture in the camera is arranged
The assessment device of joining quality.Optionally, the assessment device of the Panorama Mosaic quality can pass through software and/or hardware
Be implemented in combination with.
Optionally, the altimetric image to be evaluated can be panoramic picture 13 or panoramic picture 14 in Fig. 1.
For example, reference picture is panoramic picture 14 in Fig. 1 when altimetric image to be evaluated is panoramic picture 13 in Fig. 1.For example,
When altimetric image to be evaluated is panoramic picture 14 in Fig. 1, reference picture is panoramic picture 13 in Fig. 1.
Optionally, dead zone information can according at least two first image mosaics at altimetric image to be evaluated during generate
Blind area image size.
Optionally, pixel information can be obtained by following feasible formula 1:
Wherein, A is pixel information, and π is nature pi, and σ and μ are parameter preset, and x is predeterminable area in panoramic picture
In image.
Optionally, the predeterminable area can be the image presetting splicing regions or presetting in non-splicing regions.Specifically
Refer to Fig. 4.
It should be noted that obtaining image method to be evaluated according at least two first images and according at least two
Second image obtains reference picture method, can refer to Fig. 4.
Optionally, the method for obtaining frequency information may include: by the image down in predeterminable area to 8*8 pixel
Third image is converted to gray level image after obtaining third image by the third image of composition, is carried out to gray level image discrete remaining
String converts (Discrete Cosine Transform, DCT) processing, obtains the first matrix, first matrix includes 32*32
A element determines that (second matrix is that the 8*8 element in the upper left corner in the first matrix forms to the second matrix according to the first matrix
, second matrix shows the low-limit frequency in image), after obtaining the second matrix, 8*8 are determined in the second matrix
The average value of element sets 1 for the element, when the second square when the element in the second matrix is more than or equal to average value
When element in battle array is less than average value, 0 is set by the element, finally, by the second matrix by rows including 8*8 0 or 1
Expansion obtains frequency information to realize.
It should be noted that fade chart can be generated during at least two first images of splicing obtain altimetric image to be evaluated
Picture, the blind area image are the image for not including, and including in altimetric image to be evaluated at least two first images.
S202: according to the first attribute information and the second attribute information, the joining quality of altimetric image to be evaluated is determined.
Optionally, first eigenvector can be determined, according to first according to the first attribute information and the second attribute information
Feature vector and presetting module determine the joining quality of altimetric image to be evaluated.
Optionally, the first eigenvector includes n element.For example, the value of n can be 5 etc..
It should be noted that the presetting module is obtained according to disclosed data set, the disclosed data set is
The data set about the evaluation of panoramic picture evaluation quality of the laboratory BJ University of Aeronautics & Astronautics CVTEAM publication in 2019.
In the assessment method of image mosaic quality provided by the invention, obtain altimetric image to be evaluated the first attribute information and
Second attribute information of reference picture, the first attribute information include the dead zone information of altimetric image to be evaluated, pixel information and frequency
Information, the second attribute information include the pixel information and frequency information of reference picture, and altimetric image to be evaluated is splicing at least two
What the first image obtained, reference picture splices at least two second images and obtains, and is belonged to according to the first attribute information and second
Property information, determines the joining quality of altimetric image to be evaluated.In above process, according to the first attribute information and the second attribute information,
The joining quality for determining altimetric image to be evaluated improves the accuracy tested and assessed to the joining quality of image to be evaluated.
Fig. 3 is the flow diagram two of the assessment method of image mosaic quality provided in an embodiment of the present invention.Refer to figure
3, the assessment method of image mosaic quality includes:
S301: obtaining the first attribute information of altimetric image to be evaluated and the second attribute information of reference picture, and described first belongs to
Property information includes dead zone information, pixel information and the frequency information of the altimetric image to be evaluated, and second attribute information includes
The pixel information and frequency information of the reference picture, the altimetric image to be evaluated are that at least two first images of splicing obtain
, the reference picture splices at least two second images and obtains.
It should be noted that the execution method of S301 and S201 execution method are identical, herein, not in the execution for repeating S301
Process.
S302: according to the dead zone information in the first attribute information, the blind area feature of altimetric image to be evaluated is determined.
Optionally, when the dead zone information is the size of blind area image, the blind area feature of the altimetric image to be evaluated
Blind can be the ratio of the area s2 of the area s1 and altimetric image to be evaluated of fade chart picture.
Such as: the area s1 of blind area image is 4cm2, altimetric image to be evaluated area s2 be 40cm2, then blind area feature Blind
It can be 0.1, i.e. Blind=s1/s2=4/40=0.1.
S303: according to the pixel information of pixel information and the second attribute information in the first attribute information, determine to
The histogram statistical features of evaluation and test image and sparse reconstruction features.
Optionally, the pixel information in the first attribute information is that the splicing regions (area in Fig. 4 is preset in altimetric image to be evaluated
Domain 1) interior image pixel information.
Optionally, the pixel information in the second attribute information is to preset the non-splicing regions (area in Fig. 4 in reference picture
Domain 3) interior image pixel information.
Optionally, presetting splicing regions and presetting non-splicing regions human eye gaze frequency range -70 can spend extremely according to
70 degree determining.Wherein, the gaze frequency range of -70 degree to 70 degree indicates default splicing regions and presets non-splicing regions
Height.
It should be noted that the pixel information includes the number of pixel and the gray value of pixel.
It optionally, can be according to image in the pixel information and non-splicing regions of image in above-mentioned default splicing regions
Pixel information, determine altimetric image to be evaluated histogram statistical features and sparse reconstruction features.
Optionally, image in default splicing regions can be handled according to the formula 1 in S201, obtains first and belongs to
Pixel information A in property information0.It can be handled according to the formula 1 in S201 image in non-splicing regions is preset,
Obtain the pixel information A in the second attribute information1。
Optionally, by pixel information A0Statistics with histogram is carried out according to preset multiple gray value intervals, is determined each
The number of pixel in gray value interval determines primary vector m according to the number of pixel in each gray value interval0, described
The number of element is equal to the number of gray value interval in primary vector.
For example, 0~225 gray value is divided into 32 sections, then the first gray value area is [1,15], the second gray value
Area is [16,30], wherein the first element in primary vector is the number for the pixel that gray value belongs to the first gray value.
Optionally, can according to pass through pixel information A0Determine primary vector m0Method, pass through pixel information A1
Determine secondary vector m1。
Optionally, the histogram statistical features His of altimetric image to be evaluated can be determined according to following feasible formula 2:
Wherein, q1 is primary vector m0In element number, q2 be secondary vector m1In element number.For first to
Measure m0In i-th of element,For secondary vector m1In j-th of element.
Optionally, according to primary vector m0With secondary vector m1, by following formula 3, determine the sparse of altimetric image to be evaluated
Encode X*:
Wherein, T is basal orientation moment matrix, can be obtained by study, λ1For parameter preset.
Optionally, according to sparse coding X*, by following formula 4, determine nonnegative real number diagonal matrix Σ:
X*=U Σ V*Formula 4
Wherein, U is unitary matrice, V*For unitary matrice, the U and V*In element it is different.
Optionally, sparse reconstruction features can be determined by following formula 5 according to nonnegative real number diagonal matrix Σ
Sparse:
Sparse=∑ num (Σ) formula 5
Wherein, sparse reconstruction features Sparse is the element (i.e. the number of singular value) on Σ diagonal line.
S304: it according to the frequency information of frequency information and the second attribute information in the first attribute information, determines to be evaluated
The perceived hash characteristics of image.
Optionally, it according to the method for obtaining frequency information in S201, can be preset in splicing regions in image to be evaluated
Image handled, obtain the first attribute information in frequency information H1, can be to presetting non-splicing regions in reference picture
In image handled, obtain the second attribute information in frequency information H2。
It optionally, can be according to frequency information H1With frequency information H2, determine the perceived hash characteristics of altimetric image to be evaluated
PHash。
S305: according to altimetric image to be evaluated and at least two first images, the color difference feature of altimetric image to be evaluated is determined.
Optionally, the color difference feature of altimetric image to be evaluated can be determined by following feasible method: according to altimetric image to be evaluated
With at least two first images, using Scale invariant features transform (Scale-invariant feature transform,
SIFT), determine in altimetric image to be evaluated and at least two first images and N is led to according to pixel reference point to N related like vegetarian refreshments
Following feasible formula 7 is crossed, the color difference feature Color of altimetric image to be evaluated is obtained:
Wherein, C is default channel number, and value can be 3, λ2For parameter preset, SijIt is i-th of altimetric image to be evaluated
Related like gray value of the vegetarian refreshments in j-th of channel, RijFor i-th at least two first images related like vegetarian refreshments
Gray value in j channel.
S306: according to blind area feature, histogram statistical features, sparse reconstruction features, perceived hash characteristics and color difference feature,
Determine described eigenvector.
Optionally, by blind area feature Blind, histogram statistical features His, sparse reconstruction features Sparse, perceptual hash
Feature PHash and color difference feature Color are combined, and determine described eigenvector.
For example, feature vector R=[His, Sparse, PHash, Color, Blind].
S307: obtaining preset model, and the preset model learns to obtain according to multiple groups sample image, every group of sampled images
This includes the feature vector sample joining quality corresponding with this image of sample image.
In a kind of possible embodiment, the preset model can be indicated in the form of following formula 8:
Wherein,For the output valve of the preset model, β is the parameter matrix of the preset model, and x is the default mould
The feature vector to be entered of type.
It should be noted that in above-mentioned formula 8, when preset model receives some feature vector, the default mould
The output valve of typeIt is less than preset threshold with the difference of the quality of the panoramic picture of user's observation, the preset threshold can be
0.1。
Optionally, parameter matrix β following feasible formula 9 can be determined according to:
Wherein, parameter matrix β can be learnt to obtain according to multiple groups sample image, and X is the spy of multiple groups sample image
The matrix of vector composition is levied, Ω is the covariance matrix of residual error, and H is the square of the corresponding sample joining quality composition of sample image
Battle array.
S308: by preset model, being handled feature vector, obtains the joining quality of altimetric image to be evaluated.
Optionally, blind area feature Blind, histogram statistical features His, sparse reconstruction features are also obtained in the present invention
The similarity of any one feature and real features " 1 " in Sparse, perceived hash characteristics PHash and color difference feature Color,
As shown in Figure 1.
Table 1
| Index | His | PHash | Sparse | Color | Blind | Fusion feature |
| Similarity | 0.825 | 0.841 | 0.794 | 0.803 | 0.660 | 0.948 |
It should be noted that fusion feature is the combination of above-mentioned five features, according to table 1, the combination of 5 features is obtained
The similarity taken is maximum, that is, the joining quality using the altimetric image to be evaluated of the combination acquisition of five features is best, to figure to be evaluated
The assessment of the joining quality of picture is more acurrate.
For example, the method provided according to the present invention, tests and assesses to the joining quality of multiple altimetric images to be evaluated, acquisition it is more
Kind evaluating result is ranked up the joining quality of multiple altimetric images to be evaluated according to the size of a variety of evaluating results, obtains splicing
Quality ordering vector (such as: the ordering vector of joining quality be [1,2,4,3,5], wherein " 1 " refers to the first picture).
Optionally, after the ordering vector for obtaining joining quality, the spelling of human eye observation's ordering vector and various methods is calculated
Connect five indexs of the ordering vector of quality: cosine similarity (Cosine Similarity, CS), Pearson came grade phase relation
Number (Pearson Rank Correlation Coefcient, PRCC), Spearman rank correlation coefficient (Spearman ' s
Rank Order Correlation Coefcient, ROCC), Kendall's tau coefficient (Kendall Rank
Correlation Coefcient, KRCC), root mean square error (Root Mean Square Error, RMSE), it is each to determine
The quality of the performance of kind method.Wherein, CS, PRCC, SROCC, KRCC are higher, and RMSE is lower, then explanation is obtained by this method
Joining quality it is better.Optionally, the present invention provides utilize method of the invention and existing method (average subjective scores
(Mean of Score, MOS), mean square error (Mean Squared Error, MSE), Y-PSNR (Peak Signal
To Noise Ratio, PSNR), structural similarity (structural similarity index, SSIM), perceptual map image quality
Measure evaluator (Perception based Image Quality Evaluator, PIQE), convolutional neural networks
(Convolutional Neural Network, CNN)) above-mentioned 5 indexs are obtained, as shown in table 2 below:
Table 2
| Index | MOS | MSE | PSNR | SSIM | PIQE | CNN | Method of the invention |
| CS | 1.000 | 0.798 | 0.832 | 0.782 | 0.895 | 0.832 | 0.948 |
| PRCC | 1.000 | 0.012 | 0.158 | 0.089 | 0.476 | 0.158 | 0.737 |
| SROCC | 1.000 | 0.012 | 0.158 | 0.089 | 0.475 | 0.158 | 0.737 |
| KRCC | 1.000 | 0.024 | 0.143 | 0.071 | 0.365 | 0.103 | 0.602 |
| RMSE | 0.000 | 2.807 | 2.558 | 2.938 | 2.003 | 2.567 | 1.389 |
As shown in Table 2,5 indexs got using method shown in the present invention are more preferable relative to existing method, that is,
It is tested and assessed using joining quality of the method shown in the present invention to image to be evaluated, available more accurate result.
In embodiments of the present invention, be related to local feature (His, Sparse, PHash) and global characteristics (Color,
Blind feature representation (referring to formula 1 to 9)) promotes the accuracy rate (such as Fig. 2) about Panorama Mosaic quality evaluation.
Fig. 4 is the splicing of altimetric image to be evaluated and reference picture provided in an embodiment of the present invention.On the basis of Fig. 1,
Refer to Fig. 4, including image A to be tested and assessed, reference picture B and reference picture C, wherein image A to be tested and assessed includes default splice region
Non- splicing regions 5 and default splicing regions 2 are preset in domain 1, and reference picture B includes default splicing regions 6, presets non-splicing regions 3
With default splicing regions 7, reference picture C includes default splicing regions 8, presets non-splicing regions 4 and default splicing regions 9.
Optionally, image A to be tested and assessed can be the panoramic picture 13 in Fig. 1, and reference picture B can be the panorama in Fig. 1
Image 14.Wherein, two third images 105 and third image 106 splice and to obtain according to reference picture C.When 10, camera
When coordinate center O point, 11 face of camera, 270 degree of directions (12 face of camera, 90 degree of directions), camera 11 can be acquired
Third image 105 is obtained, camera 12 can collect third image 106.
It in practical applications, can be default splice region in above-mentioned Fig. 2 and Fig. 3 default splicing regions as described in the examples
Domain 1 and/or default splicing regions 2.
Optionally, when default splicing regions can be to preset splicing regions 1, presetting non-splicing regions can be to preset non-spelling
Connect region 3.
Optionally, when default splicing regions can be to preset splicing regions 2, presetting non-splicing regions can be to preset non-spelling
Connect region 4.
Optionally, when default splicing regions can be to preset splicing regions 1 and 2, it can be default for presetting non-splicing regions
Non- splicing regions 3 and 4.
Fig. 5 is the structural schematic diagram of the assessment device of image mosaic quality provided in an embodiment of the present invention.Fig. 5 is referred to,
The assessment device of image mosaic quality includes obtaining module 51 and determining module 52, wherein
The acquisition module 51 is used for, and obtains the first attribute information of altimetric image to be evaluated and the second attribute letter of reference picture
Breath, first attribute information include dead zone information, pixel information and the frequency information of the altimetric image to be evaluated, and described second
Attribute information includes the pixel information and frequency information of the reference picture, and the altimetric image to be evaluated is splicing at least two the
What one image obtained, the reference picture splices at least two second images and obtains；
The determining module 52 is used for, according to first attribute information and second attribute information, determine it is described to
Evaluate and test the joining quality of image.
The assessment device for the image mosaic quality that inventive embodiments provide can execute skill shown in above method embodiment
Art scheme, realization principle and beneficial effect are similar, are no longer repeated herein.
In a kind of possible embodiment, the determining module 2 is specifically used for:
According to first attribute information and second attribute information, the feature vector of the altimetric image to be evaluated is determined；
Described eigenvector is handled, the joining quality of the altimetric image to be evaluated is obtained.
In alternatively possible embodiment, the determining module 52 is specifically used for:
According to the dead zone information in first attribute information, the blind area feature of the altimetric image to be evaluated is determined；
According to the pixel information of pixel information and the second attribute information in first attribute information, determine described in
The histogram statistical features of altimetric image to be evaluated and sparse reconstruction features；
According to the frequency information of frequency information and the second attribute information in first attribute information, determine described to be evaluated
The perceived hash characteristics of altimetric image；
According to the altimetric image to be evaluated and at least two first image, determine that the color difference of the altimetric image to be evaluated is special
Sign；
According to the blind area feature, histogram statistical features, sparse reconstruction features, perceived hash characteristics and color difference feature,
Determine described eigenvector.
In alternatively possible embodiment, the determining module 52 is specifically used for:
Preset model is obtained, the preset model learns to obtain according to multiple groups sample image, every group of sampled images this packet
Include the feature vector and the corresponding sample joining quality of the sample image of sample image；
By the preset model, described eigenvector is handled, obtains the joining quality of the altimetric image to be evaluated.
In alternatively possible embodiment, the preset model are as follows:
Wherein,For the output valve of the preset model, β is the parameter matrix of the preset model, and x is the default mould
The feature vector to be entered of type.
The assessment device for the image mosaic quality that inventive embodiments provide can execute skill shown in above method embodiment
Art scheme, realization principle and beneficial effect are similar, are no longer repeated herein.
The embodiment of the present invention provides a kind of assessment device of image mosaic quality, comprising: processor, the processor with deposit
Reservoir coupling；
The memory is used for, and stores computer program；
The processor is used for, and executes the computer program stored in the memory, so that described image splices matter
The assessment device of amount executes the assessment method of image mosaic quality described in above-mentioned any means embodiment.
The embodiment of the present invention provides a kind of readable storage medium storing program for executing, including program or instruction, when described program or instruction are being counted
When running on calculation machine, the assessment method of the image mosaic quality as described in above-mentioned any means embodiment is performed.
Those of ordinary skill in the art will appreciate that: realize that all or part of the steps of above-mentioned each method embodiment can lead to
The relevant hardware of program instruction is crossed to complete.Program above-mentioned can be stored in a computer readable storage medium.The journey
When being executed, execution includes the steps that above-mentioned each method embodiment to sequence；And storage medium above-mentioned include: ROM, RAM, magnetic disk or
The various media that can store program code such as person's CD.
Finally, it should be noted that the above various embodiments is only to illustrate the technical solution of the embodiment of the present invention, rather than to it
Limitation；Although the embodiment of the present invention is described in detail referring to foregoing embodiments, those skilled in the art
It is understood that it is still possible to modify the technical solutions described in the foregoing embodiments, either to part of or
All technical features are equivalently replaced；And these are modified or replaceed, it does not separate the essence of the corresponding technical solution this hair
The range of bright example scheme.
Claims (10)
1. a kind of assessment method of image mosaic quality characterized by comprising
The first attribute information of altimetric image to be evaluated and the second attribute information of reference picture are obtained, first attribute information includes
Dead zone information, pixel information and the frequency information of the altimetric image to be evaluated, second attribute information include described with reference to figure
The pixel information and frequency information of picture, the altimetric image to be evaluated splice at least two first images and obtain, the reference
Image splices at least two second images and obtains；
According to first attribute information and second attribute information, the joining quality of the altimetric image to be evaluated is determined.
2. the method according to claim 1, wherein described belong to according to first attribute information with described second
Property information, determines the joining quality of the altimetric image to be evaluated, comprising:
According to first attribute information and second attribute information, the feature vector of the altimetric image to be evaluated is determined；
Described eigenvector is handled, the joining quality of the altimetric image to be evaluated is obtained.
3. according to the method described in claim 2, it is characterized in that, described belong to according to first attribute information with described second
Property information, determines the feature vector of the altimetric image to be evaluated, comprising:
According to the dead zone information in first attribute information, the blind area feature of the altimetric image to be evaluated is determined；
According to the pixel information of pixel information and the second attribute information in first attribute information, determine described to be evaluated
The histogram statistical features of altimetric image and sparse reconstruction features；
According to the frequency information of frequency information and the second attribute information in first attribute information, the mapping to be evaluated is determined
The perceived hash characteristics of picture；
According to the altimetric image to be evaluated and at least two first image, the color difference feature of the altimetric image to be evaluated is determined；
According to the blind area feature, histogram statistical features, sparse reconstruction features, perceived hash characteristics and color difference feature, determine
Described eigenvector.
4. according to the method described in claim 3, obtaining described it is characterized in that, described handle described eigenvector
The joining quality of altimetric image to be evaluated, comprising:
Preset model is obtained, the preset model learns to obtain according to multiple groups sample image, and every group of sampled images originally include sample
The feature vector of this image and the corresponding sample joining quality of the sample image；
By the preset model, described eigenvector is handled, obtains the joining quality of the altimetric image to be evaluated.
5. according to the method described in claim 4, it is characterized in that, the preset model are as follows:
Wherein,For the output valve of the preset model, β is the parameter matrix of the preset model, and x is the preset model
Feature vector to be entered.
6. a kind of assessment device of image mosaic quality characterized by comprising obtain module and determining module, wherein
The acquisition module is used for, and obtains the first attribute information of altimetric image to be evaluated and the second attribute information of reference picture, institute
State the dead zone information that the first attribute information includes the altimetric image to be evaluated, pixel information and frequency information, second attribute
Information includes the pixel information and frequency information of the reference picture, and the altimetric image to be evaluated is at least two first figures of splicing
As obtaining, the reference picture splices at least two second images and obtains；
The determining module is used for, and according to first attribute information and second attribute information, determines the mapping to be evaluated
The joining quality of picture.
7. device according to claim 6, which is characterized in that the determining module is specifically used for:
According to first attribute information and second attribute information, the feature vector of the altimetric image to be evaluated is determined；
Described eigenvector is handled, the joining quality of the altimetric image to be evaluated is obtained.
8. device according to claim 7, which is characterized in that the determining module is specifically used for:
According to the dead zone information in first attribute information, the blind area feature of the altimetric image to be evaluated is determined；
According to the pixel information of pixel information and the second attribute information in first attribute information, determine described to be evaluated
The histogram statistical features of altimetric image and sparse reconstruction features；
According to the frequency information of frequency information and the second attribute information in first attribute information, the mapping to be evaluated is determined
The perceived hash characteristics of picture；
According to the altimetric image to be evaluated and at least two first image, the color difference feature of the altimetric image to be evaluated is determined；
According to the blind area feature, histogram statistical features, sparse reconstruction features, perceived hash characteristics and color difference feature, determine
Described eigenvector.
9. device according to claim 8, which is characterized in that the determining module is specifically used for:
Preset model is obtained, the preset model learns to obtain according to multiple groups sample image, and every group of sampled images originally include sample
The feature vector of this image and the corresponding sample joining quality of the sample image；
By the preset model, described eigenvector is handled, obtains the joining quality of the altimetric image to be evaluated.
10. device according to claim 9, which is characterized in that the preset model are as follows:
Wherein,For the output valve of the preset model, β is the parameter matrix of the preset model, and x is the preset model
Feature vector to be entered.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910402871.9A CN110276744B (en) | 2019-05-15 | 2019-05-15 | Image splicing quality evaluation method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910402871.9A CN110276744B (en) | 2019-05-15 | 2019-05-15 | Image splicing quality evaluation method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110276744A true CN110276744A (en) | 2019-09-24 |
| CN110276744B CN110276744B (en) | 2021-10-26 |
Family
ID=67959919
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910402871.9A Active CN110276744B (en) | 2019-05-15 | 2019-05-15 | Image splicing quality evaluation method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110276744B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110838119A (en) * | 2019-11-15 | 2020-02-25 | 珠海全志科技股份有限公司 | Human face image quality evaluation method, computer device and computer readable storage medium |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8213737B2 (en) * | 2007-06-21 | 2012-07-03 | DigitalOptics Corporation Europe Limited | Digital image enhancement with reference images |
| CN104408716A (en) * | 2014-11-24 | 2015-03-11 | 宁波大学 | Three-dimensional image quality objective evaluation method based on visual fidelity |
| EP3142045A1 (en) * | 2015-09-09 | 2017-03-15 | Accenture Global Services Limited | Predicting accuracy of object recognition in a stitched image |
| CN106920215A (en) * | 2017-03-06 | 2017-07-04 | 长沙全度影像科技有限公司 | A kind of detection method of panoramic picture registration effect |
| CN107610093A (en) * | 2017-08-02 | 2018-01-19 | 西安理工大学 | Full-reference image quality evaluating method based on similarity feature fusion |
| CN107705286A (en) * | 2017-08-23 | 2018-02-16 | 南京航空航天大学 | A kind of color image quality integrated evaluating method |
| CN108074237A (en) * | 2017-12-28 | 2018-05-25 | 广东欧珀移动通信有限公司 | Approach for detecting image sharpness, device, storage medium and electronic equipment |
| CN108074241A (en) * | 2018-01-16 | 2018-05-25 | 深圳大学 | Quality score method, apparatus, terminal and the storage medium of target image |
| CN108596890A (en) * | 2018-04-20 | 2018-09-28 | 浙江科技学院 | A kind of full reference picture assessment method for encoding quality that view-based access control model measured rate adaptively merges |
| CN108830823A (en) * | 2018-03-14 | 2018-11-16 | 西安理工大学 | The full-reference image quality evaluating method of frequency-domain analysis is combined based on airspace |
-
2019
- 2019-05-15 CN CN201910402871.9A patent/CN110276744B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8213737B2 (en) * | 2007-06-21 | 2012-07-03 | DigitalOptics Corporation Europe Limited | Digital image enhancement with reference images |
| CN104408716A (en) * | 2014-11-24 | 2015-03-11 | 宁波大学 | Three-dimensional image quality objective evaluation method based on visual fidelity |
| EP3142045A1 (en) * | 2015-09-09 | 2017-03-15 | Accenture Global Services Limited | Predicting accuracy of object recognition in a stitched image |
| CN106920215A (en) * | 2017-03-06 | 2017-07-04 | 长沙全度影像科技有限公司 | A kind of detection method of panoramic picture registration effect |
| CN107610093A (en) * | 2017-08-02 | 2018-01-19 | 西安理工大学 | Full-reference image quality evaluating method based on similarity feature fusion |
| CN107705286A (en) * | 2017-08-23 | 2018-02-16 | 南京航空航天大学 | A kind of color image quality integrated evaluating method |
| CN108074237A (en) * | 2017-12-28 | 2018-05-25 | 广东欧珀移动通信有限公司 | Approach for detecting image sharpness, device, storage medium and electronic equipment |
| CN108074241A (en) * | 2018-01-16 | 2018-05-25 | 深圳大学 | Quality score method, apparatus, terminal and the storage medium of target image |
| CN108830823A (en) * | 2018-03-14 | 2018-11-16 | 西安理工大学 | The full-reference image quality evaluating method of frequency-domain analysis is combined based on airspace |
| CN108596890A (en) * | 2018-04-20 | 2018-09-28 | 浙江科技学院 | A kind of full reference picture assessment method for encoding quality that view-based access control model measured rate adaptively merges |
Non-Patent Citations (3)
| Title |
|---|
| KAIWEN YU 等: ""Image Quality Assessment for Omnidirectional Cross-reference Stitching"", 《ARXIV》 * |
| 宫妍: ""全景图像拼接关键技术研究"", 《中国博士学位论文全文数据库-信息科技辑》 * |
| 钱鑫大: ""基于双目视觉的图像拼接技术研究"", 《中国优秀硕士学位论文全文数据库-信息科技辑》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110838119A (en) * | 2019-11-15 | 2020-02-25 | 珠海全志科技股份有限公司 | Human face image quality evaluation method, computer device and computer readable storage medium |
| CN110838119B (en) * | 2019-11-15 | 2022-03-04 | 珠海全志科技股份有限公司 | Human face image quality evaluation method, computer device and computer readable storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110276744B (en) | 2021-10-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106920232B (en) | Gradient similarity graph image quality evaluation method and system based on conspicuousness detection | |
| CN108428227B (en) | No-reference image quality evaluation method based on full convolution neural network | |
| Liu et al. | Image retargeting quality assessment | |
| WO2020010785A1 (en) | Classroom teaching cognitive load measuring system | |
| CN102169576B (en) | Quantified evaluation method of image mosaic algorithms | |
| Zhang et al. | A no-reference evaluation metric for low-light image enhancement | |
| DE112016001830T5 (en) | Discovering companies from images | |
| Tian et al. | A multi-order derivative feature-based quality assessment model for light field image | |
| Bellavia et al. | Dissecting and reassembling color correction algorithms for image stitching | |
| CN104023230A (en) | Non-reference image quality evaluation method based on gradient relevance | |
| CN107578395B (en) | Image quality evaluation method based on visual perception | |
| CN106504207B (en) | A kind of image processing method | |
| Cheng et al. | Screen content video quality assessment: Subjective and objective study | |
| CN113610046B (en) | Behavior recognition method based on depth video linkage characteristics | |
| CN106023214B (en) | Image quality evaluating method and system based on central fovea view gradient-structure similitude | |
| WO2022116104A1 (en) | Image processing method and apparatus, and device and storage medium | |
| LI et al. | Recent advances and challenges in video quality assessment | |
| Li et al. | Local and global sparse representation for no-reference quality assessment of stereoscopic images | |
| CN114677644A (en) | Student seating distribution identification method and system based on classroom monitoring video | |
| CN110276744A (en) | The assessment method and device of image mosaic quality | |
| Chen et al. | A novel face super resolution approach for noisy images using contour feature and standard deviation prior | |
| Zeng et al. | Screen content video quality assessment model using hybrid spatiotemporal features | |
| Liu et al. | Spatio-temporal interactive laws feature correlation method to video quality assessment | |
| CN112954313A (en) | Method for calculating perception quality of panoramic image | |
| Yang et al. | EHNQ: Subjective and objective quality evaluation of enhanced night-time images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |