CN109509201A - A kind of SAR image quality evaluating method and device - Google Patents
A kind of SAR image quality evaluating method and device Download PDFInfo
- Publication number
- CN109509201A CN109509201A CN201910006999.3A CN201910006999A CN109509201A CN 109509201 A CN109509201 A CN 109509201A CN 201910006999 A CN201910006999 A CN 201910006999A CN 109509201 A CN109509201 A CN 109509201A
- Authority
- CN
- China
- Prior art keywords
- region
- image
- evaluated
- similarity
- original 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.)
- Withdrawn
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10032—Satellite or aerial image; Remote sensing
- G06T2207/10044—Radar image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30168—Image quality inspection
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Quality & Reliability (AREA)
- Radar Systems Or Details Thereof (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to a kind of SAR image quality evaluating method and devices, one embodiment of the method includes: that original image and image to be evaluated are divided into multiple regions respectively with identical division mode, and each corresponding region of original image and image to be evaluated is a region pair；Obtain the index of similarity between each region centering original image region and image-region to be evaluated；For each region pair, the scattering center feature in original image region and image-region to be evaluated is obtained respectively and determines the similarity of the two, the weight factor in the region pair is obtained using the similarity；The global index of similarity between original image and image to be evaluated is determined according to the index of similarity in each region pair and weight factor, using the overall situation index of similarity as the evaluation index of picture quality to be evaluated.The embodiment is capable of providing the target SAR image quality evaluating method for meeting human visual system.
Description
Technical field
The present invention relates to target classification and identification technical field more particularly to a kind of SAR image quality evaluating methods and dress
It sets.
Background technique
As various countries deeply develop marine resources, marine environment also becomes increasingly complicated.Either military field
Investigation and strike or the exploitation and fishing of civil field, the classification of sea-surface target and identification are to defending state sovereignty, remains extra large
Foreign order plays a crucial role.
Synthetic aperture radar (SAR) is because having round-the-clock, round-the-clock imaging not available for optical remote sensing imaging system
Ability and be widely used in sea-surface target investigation with identification technology field, SAR imaging process mainly reflect target electromagnetism dissipate
Penetrate characteristic, and also to allow in SAR image Ship Target to have higher quick for the characteristics such as the metal material of Ship Target, up rightness structure
Perception.Therefore, Ship Target SAR image has high application value in the Classification and Identification of sea-surface target.
But due to the difference of imaging mode, there are biggish differences with remote sensing image for SAR image.Sea at present
The major way of SAR image detection and the identification of target is man-machine interactive, i.e., gives otherwise first with automatic detection with knowledge
Certain initial screening and anticipation out, recycle the mode of manual reading of drawings to be confirmed and identified.The detection of this mode identifies quasi-
True property and efficiency and SAR image quality level are closely related.Therefore, the naval vessel mesh for meeting human visual system (HVS) is found
Marking SAR image quality evaluating method has very important application value.
Summary of the invention
The technical problem to be solved by the present invention is to how provide the target SAR image quality evaluation for meeting human visual system
Method.
In order to solve the above-mentioned technical problem, in one aspect, the present invention provides a kind of SAR image quality evaluating methods.
The SAR image quality evaluating method of the embodiment of the present invention can be used for similar to original image according to image to be evaluated
Property judges the quality of image to be evaluated, wherein image to be evaluated and original image are all SAR image；The described method includes: with phase
With division mode original image and image to be evaluated are divided into multiple regions respectively, original image and image to be evaluated it is every
One corresponding region is a region pair；It obtains similar between each region centering original image region and image-region to be evaluated
Sex index；For each region pair, the scattering center feature in original image region and image-region to be evaluated and true is obtained respectively
Both fixed similarity, the weight factor in the region pair is obtained using the similarity；Referred to according to the similitude in each region pair
Several and weight factor determines the global index of similarity between original image and image to be evaluated, which is made
For the evaluation index of picture quality to be evaluated.
Preferably, the scattering center feature for obtaining original image region and image-region to be evaluated respectively is specific to wrap
It includes: for original image region or image-region to be evaluated, obtaining the scattering center of the regions scatter maximum intensity；By the scattering
The vector of multiple attribute scattering center model parameters composition at center is determined as the scattering center feature in the region.
Preferably, the similarity is cosine similarity；And the power that the region pair is obtained using the similarity
Repeated factor specifically includes: executing edge detection to the original image region of the region centering, is obtained according to edge pixel point quantity
The impact factor in the original image region, using the impact factor as the impact factor in the region pair；In conjunction with the similarity and
The impact factor determines the weight factor in each region pair.
Preferably, the division mode is half overlapping division mode, and the index of similarity is structural similarity index；With
And the method further includes: multiple regions are obtained to later, according to each region centering original image area dividing image
The pixel value standard deviation of domain or image-region to be evaluated judge the region to for simple region to or complex region pair.
Preferably, the similitude obtained between each region centering original image region and image-region to be evaluated refers to
Number, specifically includes: for each complex region pair, respectively by the original image region of the region centering and image-region to be evaluated
Carry out second level wavelet decomposition；The region centering original image region and image-region to be evaluated are calculated using obtained low frequency sub-band
Brightness fiducial value and contrast fiducial value, using obtained high-frequency sub-band calculate the region centering original image region with it is to be evaluated
The structure fiducial value of valence image-region；In conjunction with the brightness fiducial value, the contrast fiducial value and the structure fiducial value meter
Calculate the structural similarity index between the region centering original image region and image-region to be evaluated.
On the other hand, the embodiment of the present invention provides a kind of SAR image quality evaluation device.
The SAR image quality evaluation device of the embodiment of the present invention can be used for similar to original image according to image to be evaluated
Property judges the quality of image to be evaluated, wherein image to be evaluated and original image are all SAR image；Described device can include: phase
Like sex index computing unit, for original image and image to be evaluated to be divided into multiple areas respectively with identical division mode
Each corresponding region of domain, original image and image to be evaluated is a region pair；Obtain each region centering original image area
Index of similarity between domain and image-region to be evaluated；Weight Acquisition unit, is used for: for each region pair, obtaining respectively
The scattering center feature of original image region and image-region to be evaluated and the similarity for determining the two, are obtained using the similarity
Take the weight factor in the region pair；Evaluation unit, it is former for being determined according to the index of similarity and weight factor in each region pair
Global index of similarity between beginning image and image to be evaluated, using the overall situation index of similarity as picture quality to be evaluated
Evaluation index.
Preferably, the similarity is cosine similarity；And Weight Acquisition unit is further used for: for original graph
As region or image-region to be evaluated, the scattering center of the regions scatter maximum intensity is obtained；By multiple categories of the scattering center
Property scattering center model parameter composition vector be determined as the scattering center feature in the region；To the original image of the region centering
Region executes edge detection, the impact factor in the original image region is obtained according to edge pixel point quantity, by the impact factor
Impact factor as the region pair；The weight factor in each region pair is determined in conjunction with the similarity and the impact factor.
Preferably, the division mode is half overlapping division mode, and the index of similarity is structural similarity index；Institute
Stating device further comprises taxon, is used to obtain multiple regions to later, according to each region centering dividing image
The pixel value standard deviation of original image region or image-region to be evaluated judge the region to for simple region to or it is complicated
Region pair；And index of similarity computing unit is further used for: for each complex region pair, respectively by the region centering
Original image region and image-region to be evaluated carry out second level wavelet decomposition；The region pair is calculated using obtained low frequency sub-band
In original image-region and image-region to be evaluated brightness fiducial value and contrast fiducial value, utilize obtained high-frequency sub-band meter
Calculate the structure fiducial value in the region centering original image region and image-region to be evaluated；In conjunction with the brightness fiducial value, described
Contrast fiducial value and the structure fiducial value calculate between the region centering original image region and image-region to be evaluated
Structural similarity index.
It yet still another aspect, the present invention provides a kind of electronic equipment.
The electronic equipment of the embodiment of the present invention includes: one or more processors；Storage device, for storing one or more
A program, when one or more of programs are executed by one or more of processors, so that one or more of processing
Device realizes above-mentioned SAR image quality evaluating method.
In another aspect, the present invention provides a kind of computer readable storage medium.
The computer readable storage medium of the embodiment of the present invention, is stored thereon with computer program, and described program is processed
Device realizes above-mentioned SAR image quality evaluating method when executing.
Above-mentioned technical proposal of the invention has the advantages that in embodiments of the present invention, first to target SAR image into
The processing of row piecemeal, is divided into simple region and complex region；Secondly, in order to allow evaluation method to be more in line with HVS characteristic, to simple
Region carries out SSIM evaluation, carries out wavelet decomposition to complex region and is evaluated respectively different frequency range region, is then utilized
Wavelet energy coefficient is integrated to obtain the evaluation result of complex region；Again, meter is used for the limitation of electromagnetic property
The method for calculating scattering signatures information will for the method that the limitation of edge distribution information uses Canny operator edge extracting
Scattered information and marginal information are integrated to obtain weight factor.Finally, using weight factor to the evaluation result in each region
It is integrated, obtains final evaluation result.
Detailed description of the invention
Fig. 1 is the key step schematic diagram of the SAR image quality evaluating method of the embodiment of the present invention；
Fig. 2 is the model ship SAR emulation schematic diagram of the embodiment of the present invention；
Fig. 3 is the specific implementation schematic diagram of the SAR image quality evaluating method of the embodiment of the present invention；
Fig. 4 is the component part schematic diagram of the SAR image quality evaluation device of the 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 embodiments of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, ordinary skill people
Member's every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.
Fig. 1 is the key step schematic diagram of the SAR image quality evaluating method of the embodiment of the present invention.
As shown in Figure 1, following steps can be performed in the SAR image quality evaluating method of the embodiment of the present invention:
Step S101: being divided into multiple regions for original image and image to be evaluated respectively with identical division mode, former
Each corresponding region of beginning image and image to be evaluated is a region pair；Obtain each region centering original image region with to
Evaluate the index of similarity between image-region.
In embodiments of the present invention, original image and image to be evaluated are all SAR images.It is appreciated that original image refers to
Be image as reference standard, the higher image of the quality having determined may be selected as original image in practical application.
It should be noted that original image can be measuring image in actual scene, it is also possible to emulating image.
In concrete application, original image and image to be evaluated can be respectively divided using same half overlapping division mode
For multiple regions.On the one hand this division mode can carry out a degree of macro-regions division to entire area, after reduction
The blocking artifact occurred in phase evaluation, on the other hand also can reduce the complexity of algorithm, improve computational efficiency.The number of regions of division
Preset reasonable value is measured, too small can not embody of the numerical value divides effect, the excessive complexity that will increase algorithm, and
There is the meticulous problem of piecemeal.
Above-mentioned half overlapping division mode is shown in the following example: if by an abscissa in 0 to 40 ranges, ordinate exists
Rectangular image in 0 to 30 ranges is divided into 6 regions, may is that the pixel by abscissa in 0 to 20, ordinate 0 to 20
Divide in first area, abscissa is divided in 0 to 20 pixel in second area, by abscissa 20 in 10 to 30, ordinate
Divide the pixel in third region, by abscissa in 0 to 20, ordinate 10 to 30 in 0 to 20 pixel to 40, ordinate
Divide in the fourth region, abscissa is divided in 10 to 30 pixel in the 5th region, by abscissa 20 in 10 to 30, ordinate
Divide to 40, ordinate in 10 to 30 pixel in the 6th region.In the above method, the identical adjacent area of abscissa range
One semi-area of each overlapping.
In this step, the index of similarity between each region centering original image region and image-region to be evaluated can
To be structural similarity index SSIM (structural similarity index).It will be understood by those skilled in the art that obtaining
It takes SSIM to need to compare brightness, contrast and the structural information of two images, that is, utilizes brightness comparison function l (α, β), contrast
Comparison function c (α, β) and structure comparison function s (α, β) come calculate two images SSIM (α, β respectively indicate original image and
The pixel value of image to be evaluated), as is generally known in the art, the present invention is not discussed in detail above-mentioned specific calculating process.
In embodiments of the present invention, the index of similarity in each region pair can be obtained by following steps:
1. obtain multiple regions to later dividing image, according to each region centering original image region or to be evaluated
The pixel value standard deviation of image-region judge the region to for simple region to or complex region pair.Specific formula is as follows:
Wherein, k=1,2 ..., M, M are the region sum in original image or image to be evaluated, σkFor k-th of region
The standard deviation of pixel value, CkFor index of discrimination.
The complexity C of entire image is set as threshold value, by CkIt is compared with C: complexity CkThe region of >=C is complex area
Domain, CkThe region of < C is simple region, uses different evaluation methods to different regions to facilitate.It is appreciated that due to
SAR image is complex image, and above-mentioned pixel value can refer to the corresponding complex data of pixel (i.e. with the pixel of phase information
Value) mould, can also refer to the gray value be converted to SAR image after gray level image.In addition, including the region pair of simple region
For simple region pair, the region including complex region is to for complex region pair.
2., since its detailed information is relatively weak, being calculated using existing SSIM algorithm similar for simple region pair
Sex index.
3. on the one hand region therein has more detailed information, in actual application for each complex region pair
In play leading position；Another aspect human vision is higher to the degree of concern of detailed information, architectural characteristic largely according to
Rely the detailed information in complex region.Wavelet transformation has multiple dimensioned, multidirectional and local space characteristics simultaneously, therefore can make
It is evaluated with the SSIM algorithm based on wavelet transformation.
It specifically, can be first by original image region therein and image-region to be evaluated for any complex region pair
Second level wavelet decomposition is carried out, following subband sequence is obtained:
Wherein, the subband of subscript LL is low frequency sub-band, is obtained by level-one wavelet decomposition.Subscript LH1、HL1、HH1
Subband be the high-frequency sub-band obtained by level-one wavelet decomposition, subscript LH2、HL2、HH2Subband be in HH1Under subband passes through
The high-frequency sub-band that level-one wavelet decomposition (passing through two-stage wavelet decomposition altogether) obtains.Due to HH1In contained the small wavelength-division of second level
Solve subband LL2Information, therefore by LL2Removal.
Structure comparison function s (α H, β are calculated using the high-frequency sub-band in above-mentioned subband laterH ), (αH,βHIt respectively indicates
The high-frequency sub-band pixel value in original image region and image-region to be evaluated) improve SSIM algorithm medium-high frequency subband structure compare
Function, it is contemplated that the non-low frequency region distribution characteristics after wavelet decomposition defines structure comparison function at this time are as follows:
Wherein, the s function in summation sign is the structure comparison function of existing SSIM algorithm.
For the coefficient of lambda notation in above formula, can be determined by the distribution situation of each high-frequency sub-band energy.
WithCalculation method for, original image region is after 1 grade of wavelet decomposition, the wavelet coefficient total energy of LH component
Amount are as follows:
Wherein, N is original image pixel sum,For in the LH component of 1 grade of wavelet decomposition ith pixel point it is small
Wave Decomposition high frequency coefficient.The high frequency division in horizontal, vertical and diagonal three directions in every grade of wavelet decomposition is calculated separately using above formula
The gross energy of amount:
Later, image high-frequency region gross energy E is defined are as follows:
The coefficient of normalized high-frequency sub-band is obtained using E
The coefficient of remaining high-frequency sub-band can be calculated by same mode.
Finally, calculating l (α using low frequency sub-band LLLL,βLL) and c (αLL,βLL), s (α is calculated using 6 high-frequency sub-bandsH,
βH), calculated result is combined by following formula, the SSIM of complex region pair can be obtained:
SSIMcomplex(α, β)=l (αLL,βLL)·c(αLL,βLL)·s(αH,βH)
Wherein, αLL,βLLRespectively indicate the low frequency sub-band pixel value of original image region and image-region to be evaluated.
Can that is, in above-mentioned calculating process, calculate the region centering original image area using obtained low frequency sub-band
The brightness fiducial value and contrast fiducial value in domain and image-region to be evaluated calculate the region centering using obtained high-frequency sub-band
The structure fiducial value in original image region and image-region to be evaluated.In conjunction with brightness fiducial value, contrast fiducial value and structure ratio
The structural similarity index SSIM between the region centering original image region and image-region to be evaluated can be calculated compared with value.
By above step, the index of similarity in each region pair can get.
Step S102: for each region pair, respectively in the scattering of acquisition original image region and image-region to be evaluated
Heart feature and the similarity for determining the two, the weight factor in the region pair is obtained using similarity.
After obtaining index of similarity, each region is needed to determine the weight in overall image quality is evaluated
To obtain accurate evaluation result.It can be summarized according to the characteristic of HVS, in eye-observation target SAR image, Mei Gequ
The scattering signatures information that the major weight factor in domain depends on region therefore in embodiments of the present invention, can be each by determination
Its weight factor of the scattering signatures information acquisition in region pair.
Specifically, for the original image region of any region pair or image-region to be evaluated, it is strong to obtain the regions scatter
Spend maximum scattering center；The vector that multiple attribute scattering center model parameters of the scattering center form is determined as this later
The scattering center feature in region.It is understood that known attribute scattering center model (Attributed Scattering can be passed through
Center Model) and derivation algorithm, can obtain multiple attribute scattering center model parameters in a region so that construct to
Amount, calculating step can be divided into: image segmentation, scattering center classification, parameter initialization and parameter optimization, finally available to include
7 parameters including scattering center coordinate, complex magnitude, type parameter.Meanwhile judging that scattering center scattering is strong according to above-mentioned parameter
The method of degree be also it is known that specific calculating process not described in detail herein.
For each region pair, after obtaining the scattering center feature in two of them region, can calculate between the two
Similarity (such as cosine similarity) is as the region to the weight in scattering signatures dimension.It later, can be to each region pair
(for example, being normalized using the maximum similarity of each region centering) is normalized in similarity, will be similar after normalization
Spend the weight factor as region pair.
In practical application, according to the characteristic of HVS it is found that in eye-observation target SAR image, the main power in each region
Refetching certainly factor, in addition to scattering signatures information, edge feature information can also play weight when staff parses and differentiates
It acts on, therefore the weight of another dimension will be obtained by edge detection below, to optimize above-mentioned weight calculation strategy.
Preferably, in embodiments of the present invention, executing edge detection to the original image region of any region centering and (being not required to
Edge detection is carried out to image to be evaluated), the impact factor in the original image region is obtained according to edge pixel point quantity, by this
Impact factor of the impact factor as the region pair.It illustratively, can be by its original image edges of regions for a certain region pair
Pixel quantity divided by the maximum above-mentioned pixel quantity of each region centering, obtain the impact factor in the region pair.Specifically answer
In, Canny operator can be used and realize edge detection, since it belongs to known technology, details are not described herein again for specific calculating process.
After the impact factor for obtaining region pair, in combination with similarity and impact factor determine the weight in each region pair because
Son.For example, obtaining weight factor by following formula:
Wherein, k is region to serial number, mkIt is region to the similarity of k, ekIt is region to the impact factor of k,For centre
Parameter, ωkIt is region to the weight factor of k, M is region to sum.
Step S103: original image and image to be evaluated are determined according to the index of similarity in each region pair and weight factor
Between global index of similarity, using the overall situation index of similarity as the evaluation index of picture quality to be evaluated.
In this step, the weighted sum of the index of similarity of whole region pair according to the weight coefficient in region pair, can be calculated,
The weighted sum is the global index of similarity between original image and image to be evaluated, can be used as picture quality to be evaluated most
Whole evaluation index, the index is bigger, and the quality of image to be evaluated is higher.
Fig. 3 is the specific implementation schematic diagram of the SAR image quality evaluating method of the embodiment of the present invention, can be straight from Fig. 3
See the execution step and Evaluation Strategy for understanding the method for the present invention.
As it can be seen that in above-mentioned evaluation method, simple region and complex region being calculated by different modes of image
Respective similarity, and merged to obtain final quality evaluation index, this method, which considers HVS characteristic and combines, to be dissipated
Characteristic information and influence of the edge feature information to target SAR image quality evaluation are penetrated, thus, it is possible to carry out to target SAR image
Meet the accurate evaluation of HVS characteristic.Particularly, due to the characteristics such as the metal material of Ship Target, up rightness structure and the spy
Sign is more agreed with the links such as wavelet decomposition, edge detection, scattering signatures calculating in the method for the present invention, and the method for the present invention is especially suitable
Quality evaluation for Ship Target SAR image.Fig. 2 is the model ship SAR emulation schematic diagram of the embodiment of the present invention, is being schemed
In 2, abscissa is radial distance, and ordinate is lateral distance.
By test, the method for the present invention is superior to traditional MSE (mean square error), PSNR (peak value noise in following index
Than), SSIM method: 2 times of down-sampled smooth, 5 times of down-sampled smooth, 10 times of down-sampled smooth, motion blur, defocusing blurring, height
This white noise, salt-pepper noise.
In the present invention is implemented, it is further provided a kind of SAR image quality evaluation device can be used for according to image to be evaluated
The quality of image to be evaluated is judged with the similitude of original image, wherein image to be evaluated and original image are all SAR image.Institute
State can device include: index of similarity computing unit, Weight Acquisition unit and evaluation unit.
Wherein, index of similarity computing unit can be used for identical division mode respectively by original image and figure to be evaluated
As being divided into multiple regions, each corresponding region of original image and image to be evaluated is a region pair；Obtain each region
Index of similarity between centering original image region and image-region to be evaluated；Weight Acquisition unit can be used for: for each
Region pair obtains the scattering center feature in original image region and image-region to be evaluated respectively and determines the similarity of the two,
The weight factor in the region pair is obtained using the similarity；Evaluation unit can be used for the index of similarity according to each region pair
And weight factor determines the global index of similarity between original image and image to be evaluated, using the overall situation index of similarity as
The evaluation index of picture quality to be evaluated.
As a preferred embodiment, the similarity is cosine similarity；And Weight Acquisition unit can be used further
In: for original image region or image-region to be evaluated, obtain the scattering center of the regions scatter maximum intensity；By the scattering
The vector of multiple attribute scattering center model parameters composition at center is determined as the scattering center feature in the region；To the region pair
In original image region execute edge detection, the impact factor in the original image region is obtained according to edge pixel point quantity,
Using the impact factor as the impact factor in the region pair；Each region pair is determined in conjunction with the similarity and the impact factor
Weight factor.
Preferably, in embodiments of the present invention, the division mode is half overlapping division mode, the index of similarity is
Structural similarity index；Described device can further comprise taxon, be used to obtain multiple regions to it in division image
Afterwards, according to the pixel value standard deviation in each region centering original image region or image-region to be evaluated judge the region to for
Simple region to or complex region pair.
In concrete application, index of similarity computing unit can be further used for: for each complex region pair, respectively should
The original image region of region centering and image-region to be evaluated carry out second level wavelet decomposition；It is calculated using obtained low frequency sub-band
The brightness fiducial value and contrast fiducial value in the region centering original image region and image-region to be evaluated, utilize obtained height
Frequency subband calculates the structure fiducial value in the region centering original image region and image-region to be evaluated；Compare in conjunction with the brightness
Value, the contrast fiducial value and the structure fiducial value calculate the region centering original image region and image-region to be evaluated
Between structural similarity index.
In embodiments of the present invention, a kind of electronic equipment is also provided, comprising: one or more processors and storage device.
Wherein, storage device is for storing one or more programs.When one or more of programs are by one or more of processing
Device executes, so that one or more of processors realize the SAR image quality evaluating method of the embodiment of the present invention.
On the other hand, the present invention also provides a kind of computer-readable medium, which be can be
It states included in equipment described in embodiment；It is also possible to individualism, and without in the supplying equipment.Above-mentioned calculating
Machine readable medium carries one or more program, when said one or multiple programs are executed by the equipment, so that should
The step of equipment executes includes: that original image and image to be evaluated are divided into multiple regions respectively with identical division mode,
Each corresponding region of original image and image to be evaluated is a region pair；Obtain each region centering original image region with
Index of similarity between image-region to be evaluated；For each region pair, original image region and figure to be evaluated are obtained respectively
As the scattering center feature in region and the similarity of determining the two, the weight factor in the region pair is obtained using the similarity；
The global similitude between original image and image to be evaluated is determined according to the index of similarity in each region pair and weight factor
Index, using the overall situation index of similarity as the evaluation index of picture quality to be evaluated.
In conclusion in the technical solution of the embodiment of the present invention, using the characteristic of human visual system, to target
SAR image carries out the quality evaluation for having HVS characteristic, this evaluation method can the SAR image quality to target carry out meeting people
The complete of class visual characteristic refers to overall merit, provides certain reference to the image quality level of radar imaging system, and be mesh
The classification and identification for marking SAR image provide the reference frame of picture quality situation.
Finally, it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although
Present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: it still may be used
To modify the technical solutions described in the foregoing embodiments or equivalent replacement of some of the technical features；
And these are modified or replaceed, technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution spirit and
Range.
Claims (10)
1. a kind of SAR image quality evaluating method, to be evaluated for being judged according to the similitude of image to be evaluated and original image
The quality of image, wherein image to be evaluated and original image are all SAR image；It is characterized in that, which comprises
Original image and image to be evaluated are divided into multiple regions respectively with identical division mode, original image and to be evaluated
Each corresponding region of image is a region pair；Obtain each region centering original image region and image-region to be evaluated it
Between index of similarity；
For each region pair, the scattering center feature in original image region and image-region to be evaluated is obtained respectively and determines two
The similarity of person obtains the weight factor in the region pair using the similarity；
The global phase between original image and image to be evaluated is determined according to the index of similarity in each region pair and weight factor
Like sex index, using the overall situation index of similarity as the evaluation index of picture quality to be evaluated.
2. the method according to claim 1, wherein described obtain original image region and image to be evaluated respectively
The scattering center feature in region, specifically includes:
For original image region or image-region to be evaluated, the scattering center of the regions scatter maximum intensity is obtained；
The scattering center that the vector that multiple attribute scattering center model parameters of the scattering center form is determined as the region is special
Sign.
3. according to the method described in claim 2, it is characterized in that, the similarity is cosine similarity；And the utilization
The similarity obtains the weight factor in the region pair, specifically includes:
Edge detection is executed to the original image region of the region centering, which is obtained according to edge pixel point quantity
The impact factor in domain, using the impact factor as the impact factor in the region pair；
The weight factor in each region pair is determined in conjunction with the similarity and the impact factor.
4. the method according to claim 1, wherein the division mode is half overlapping division mode, the phase
It is structural similarity index like sex index；And the method further includes:
Multiple regions are obtained to later, according to each region centering original image region or image district to be evaluated dividing image
The pixel value standard deviation in domain judge the region to for simple region to or complex region pair.
5. according to the method described in claim 4, it is characterized in that, it is described obtain each region centering original image region with to
The index of similarity between image-region is evaluated, is specifically included:
For each complex region pair, the original image region of the region centering and image-region to be evaluated are subjected to second level respectively
Wavelet decomposition；
The brightness fiducial value in the region centering original image region and image-region to be evaluated is calculated using obtained low frequency sub-band
With contrast fiducial value, the region centering original image region and image-region to be evaluated are calculated using obtained high-frequency sub-band
Structure fiducial value；
The region centering original image is calculated in conjunction with the brightness fiducial value, the contrast fiducial value and the structure fiducial value
Structural similarity index between region and image-region to be evaluated.
6. a kind of SAR image quality evaluation device, to be evaluated for being judged according to the similitude of image to be evaluated and original image
The quality of image, wherein image to be evaluated and original image are all SAR image；It is characterized in that, described device includes:
Index of similarity computing unit is more for being respectively divided into original image and image to be evaluated with identical division mode
Each corresponding region of a region, original image and image to be evaluated is a region pair；Obtain each region centering original graph
As the index of similarity between region and image-region to be evaluated；
Weight Acquisition unit, is used for: for each region pair, obtaining dissipating for original image region and image-region to be evaluated respectively
It penetrates central feature and determines the similarity of the two, the weight factor in the region pair is obtained using the similarity；
Evaluation unit, for determining original image and image to be evaluated according to the index of similarity and weight factor in each region pair
Between global index of similarity, using the overall situation index of similarity as the evaluation index of picture quality to be evaluated.
7. device according to claim 6, which is characterized in that the similarity is cosine similarity；And Weight Acquisition
Unit is further used for:
For original image region or image-region to be evaluated, the scattering center of the regions scatter maximum intensity is obtained；This is dissipated
The vector for hitting multiple attribute scattering center model parameters composition of the heart is determined as the scattering center feature in the region；
Edge detection is executed to the original image region of the region centering, which is obtained according to edge pixel point quantity
The impact factor in domain, using the impact factor as the impact factor in the region pair；In conjunction with the similarity and the impact factor
Determine the weight factor in each region pair.
8. device according to claim 7, which is characterized in that the division mode is half overlapping division mode, the phase
It is structural similarity index like sex index；Described device further comprises taxon, is used to obtain in division image multiple
It is somebody's turn to do to later according to the judgement of the pixel value standard deviation of each region centering original image region or image-region to be evaluated in region
Region to for simple region to or complex region pair；And index of similarity computing unit is further used for:
For each complex region pair, the original image region of the region centering and image-region to be evaluated are subjected to second level respectively
Wavelet decomposition；The brightness ratio in region the centering original image region and image-region to be evaluated is calculated using obtained low frequency sub-band
Compared with value and contrast fiducial value, the region centering original image region and image district to be evaluated are calculated using obtained high-frequency sub-band
The structure fiducial value in domain；The region is calculated in conjunction with the brightness fiducial value, the contrast fiducial value and the structure fiducial value
Structural similarity index between centering original image region and image-region to be evaluated.
9. a kind of electronic equipment characterized by comprising
One or more processors；
Storage device, for storing one or more programs,
When one or more of programs are executed by one or more of processors, so that one or more of processors are real
Now such as method as claimed in any one of claims 1 to 5.
10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that described program is processed
Such as method as claimed in any one of claims 1 to 5 is realized when device executes.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910006999.3A CN109509201A (en) | 2019-01-04 | 2019-01-04 | A kind of SAR image quality evaluating method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910006999.3A CN109509201A (en) | 2019-01-04 | 2019-01-04 | A kind of SAR image quality evaluating method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109509201A true CN109509201A (en) | 2019-03-22 |
Family
ID=65757240
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910006999.3A Withdrawn CN109509201A (en) | 2019-01-04 | 2019-01-04 | A kind of SAR image quality evaluating method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109509201A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112785612A (en) * | 2020-08-28 | 2021-05-11 | 青岛经济技术开发区海尔热水器有限公司 | Image edge detection method based on wavelet transformation |
| CN113344843A (en) * | 2021-04-09 | 2021-09-03 | 中科创达软件股份有限公司 | Image quality evaluation method, device and system |
| CN113938671A (en) * | 2020-07-14 | 2022-01-14 | 北京灵汐科技有限公司 | Image content analysis method and device, electronic equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103871039A (en) * | 2014-03-07 | 2014-06-18 | 西安电子科技大学 | Generation method for difference chart in SAR (Synthetic Aperture Radar) image change detection |
| US20180040115A1 (en) * | 2016-08-05 | 2018-02-08 | Nuctech Company Limited | Methods and apparatuses for estimating an ambiguity of an image |
| CN108550145A (en) * | 2018-04-11 | 2018-09-18 | 北京环境特性研究所 | A kind of SAR image method for evaluating quality and device |
-
2019
- 2019-01-04 CN CN201910006999.3A patent/CN109509201A/en not_active Withdrawn
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103871039A (en) * | 2014-03-07 | 2014-06-18 | 西安电子科技大学 | Generation method for difference chart in SAR (Synthetic Aperture Radar) image change detection |
| US20180040115A1 (en) * | 2016-08-05 | 2018-02-08 | Nuctech Company Limited | Methods and apparatuses for estimating an ambiguity of an image |
| CN108550145A (en) * | 2018-04-11 | 2018-09-18 | 北京环境特性研究所 | A kind of SAR image method for evaluating quality and device |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113938671A (en) * | 2020-07-14 | 2022-01-14 | 北京灵汐科技有限公司 | Image content analysis method and device, electronic equipment and storage medium |
| CN113938671B (en) * | 2020-07-14 | 2023-05-23 | 北京灵汐科技有限公司 | Image content analysis method, image content analysis device, electronic equipment and storage medium |
| CN112785612A (en) * | 2020-08-28 | 2021-05-11 | 青岛经济技术开发区海尔热水器有限公司 | Image edge detection method based on wavelet transformation |
| CN112785612B (en) * | 2020-08-28 | 2022-09-13 | 青岛经济技术开发区海尔热水器有限公司 | Image edge detection method based on wavelet transformation |
| CN113344843A (en) * | 2021-04-09 | 2021-09-03 | 中科创达软件股份有限公司 | Image quality evaluation method, device and system |
| CN113344843B (en) * | 2021-04-09 | 2024-04-19 | 中科创达软件股份有限公司 | Image quality evaluation method, device and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Li et al. | No-reference image blur assessment based on discrete orthogonal moments | |
| Hassen et al. | Image sharpness assessment based on local phase coherence | |
| CN106504222B (en) | A kind of underwater Polarization Image Fusion system based on bionic visual mechanism | |
| Feichtenhofer et al. | A perceptual image sharpness metric based on local edge gradient analysis | |
| Liu et al. | Contrast in haze removal: Configurable contrast enhancement model based on dark channel prior | |
| Zhang et al. | Edge strength similarity for image quality assessment | |
| Zhang et al. | Visibility enhancement using an image filtering approach | |
| Khan et al. | Localization of radiance transformation for image dehazing in wavelet domain | |
| CN106530244B (en) | A kind of image enchancing method | |
| CN109509201A (en) | A kind of SAR image quality evaluating method and device | |
| CN106920232A (en) | Gradient similarity graph image quality evaluation method and system based on conspicuousness detection | |
| Gao et al. | Detail preserved single image dehazing algorithm based on airlight refinement | |
| CN105847703A (en) | Image processing method and electronic device | |
| Yeganeh et al. | Objective assessment of tone mapping algorithms | |
| CN114119436A (en) | Infrared image and visible light image fusion method and device, electronic equipment and medium | |
| CN108550145A (en) | A kind of SAR image method for evaluating quality and device | |
| Xu et al. | Deep retinex decomposition network for underwater image enhancement | |
| Junwu et al. | An infrared and visible image fusion algorithm based on LSWT-NSST | |
| Liu et al. | An efficient no-reference metric for perceived blur | |
| Guo et al. | Haze removal for single image: A comprehensive review | |
| Zhao et al. | A multi-scale U-shaped attention network-based GAN method for single image dehazing | |
| Ke et al. | Single underwater image restoration based on color correction and optimized transmission map estimation | |
| CN107451608B (en) | SAR image non-reference quality evaluation method based on multi-view amplitude statistical characteristics | |
| Yan et al. | A natural-based fusion strategy for underwater image enhancement | |
| Zhang et al. | SPCM: Image quality assessment based on symmetry phase congruency |
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 | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20190322 |
|
| WW01 | Invention patent application withdrawn after publication |