CN104537646B - The automatic MTF methods of estimation of multi-angle of remote sensing images - Google Patents
The automatic MTF methods of estimation of multi-angle of remote sensing images Download PDFInfo
- Publication number
- CN104537646B CN104537646B CN201410769046.XA CN201410769046A CN104537646B CN 104537646 B CN104537646 B CN 104537646B CN 201410769046 A CN201410769046 A CN 201410769046A CN 104537646 B CN104537646 B CN 104537646B
- Authority
- CN
- China
- Prior art keywords
- edge
- image block
- mtf
- line
- point
- 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.)
- Expired - Fee Related
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
- G06T7/12—Edge-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
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The present invention provides a kind of automatic MTF methods of estimation of the multi-angle of remote sensing images, including image block is chosen and estimates MTF two parts based on recognition status：Image block is chosen to be included：Original image is carried out into multi-angle rotary, postrotational image is obtained；Marginal point extraction is carried out respectively to postrotational image and original image, 4 breadths edge dot images are obtained；It is utilized respectively 4 width edge images to refer to, chooses the image block for estimating MTF.Included based on recognition status estimation MTF：For the image block for having obtained, make rim detection, obtain the position of marginal point, that is, obtain the sub-pix point position of marginal point；With the edge in least square fitting image block；Row interpolation is entered to the edge in the image block of fitting and average edge spread function is extracted；Simple differencing is done to average spread function, line spread function is obtained, and line spread function is fitted with Gaussian Profile；Discrete Fourier transform is carried out to the line spread function after fitting, and modulus is carried out to result, obtain final MTF sequences.
Description
Technical field
The invention belongs to remote sensing application field, in particular to a kind of automatic MTF estimation sides of the multi-angle of remote sensing images
Method.
Background technology
In remote sensing application field, modulation transfer function (MTF) is the important comprehensive evaluation index of remote sensing optical imaging system.
The height of imaging system MTF directly influences the quality of image quality：MTF is lower, the Edge texture of the remote sensing images for being obtained
Will be fuzzyyer etc. details.For satellite in orbit remote sensor, because the multiple attitude during being kept by satellite launch, track is adjusted
The influence of the severe factor of whole, cosmic space radiation, day and night temperature impact etc., the imaging performance of remote sensor can be gradually reduced, distant
Sense picture quality can gradually be deteriorated.The MTF of satellite in orbit remote sensor is detected, the application for satellite in orbit remote sensor has extremely
Important meaning.
Before remote sensing satellite heaven, in the ground experiment room stage, measured using specialized equipment and calculate its MTF, be current
Precision highest, reliability the best way.But, this method can only be carried out under laboratory conditions, it is impossible to in orbit
The MTF of satellite remote sensor is calculated.
MTF is calculated using the image-forming information of the ground target laid and the surface mark thing of selection on remote sensing images,
It is more convenient feasible for the in-orbit monitoring of remote sensing satellite.Wherein, it is representational mainly to have two class ways：One class is the U.S., method
The land-based target calibration method manually laid of the uses such as state.Another kind of is natural feature on a map method, is direct as representative with the U.S.
Using surface mark thing, the large-scale ground target of bridge, airport is such as selected, from the remote sensing images containing these targets, directly
Connect the method for calculating MTF.
Mainly employ in the prior art " sample method of comparison " and " terrestrial reference mensuration ".Sample method of comparison, is with known MTF
Sample image and satellite remote sensing images be compared and interpretation, so that it is determined that the method for the MTF of remote sensing satellite.Terrestrial reference is measured
Method, is the same, the mainly method of foreign with land-based target calibration method mentioned above.
The content of the invention
The present invention proposes that a kind of multi-angle of remote sensing images is automatic for the situation for choosing image block manually in the prior art
MTF methods of estimation.
Above-mentioned purpose of the invention realized by the technical characteristic of independent claims, and dependent claims are selecting else or have
The mode of profit develops the technical characteristic of independent claims.
To reach above-mentioned purpose, the technical solution adopted in the present invention is as follows：
A kind of automatic MTF methods of estimation of the multi-angle of remote sensing images, including image block is chosen and estimates MTF based on recognition status
Two parts, wherein：
Foregoing image block is chosen and is comprised the following steps：
Original image 1-1) is carried out into multi-angle rotary, postrotational image is obtained；
Marginal point extraction 1-2) is carried out respectively to postrotational image and original image, 4 breadths edge dot images are obtained；
1-3) it is utilized respectively 4 width edge images to refer to, chooses the image block for estimating MTF；
It is foregoing to be comprised the following steps based on recognition status estimation MTF：
2-1) for the image block for having obtained, make rim detection, obtain the position of marginal point, that is, obtain the sub- picture of marginal point
Vegetarian refreshments position；
2-2) with the edge in least square fitting image block；
2-3) enter row interpolation to the edge in the image block of fitting and extract average edge spread function；
Simple differencing 2-4) is done to average spread function, line spread function is obtained, and with Gaussian Profile to line spread function
It is fitted；
Discrete Fourier transform 2-5) is carried out to the line spread function after fitting, and modulus is carried out to result, obtain final
MTF sequences.
In further embodiment, abovementioned steps 1-1) in original image is carried out into multi-angle rotary, its anglec of rotation point
It is not：0 degree, 45 degree, 90 degree and 135 degree.
In further embodiment, abovementioned steps 1-2) in row marginal point extract, choosing Canny (Tuscany) and becoming to bring is carried out
The extraction of marginal point, obtains the edge dot image of entire image.
In further embodiment, abovementioned steps 1-3) in utilize 4 width edge images to make to be used to estimate MTF's with reference to choosing
Image block, its realization includes：
31) image block of each angle is chosen into the process that process extends as an image block width, for edge graph
Each point of picture, if the point is marginal point, starts the selection process of image block, including:
311) the height H for setting image block is 5, width w self adaptations, and width is by edge line width and edge line both sides
Peak width determines that edge line width is 1, the peak width of the edge line left and right sides its initial value W1=W2=1, both sides relation
It is W=W1+W2+1, initial image block width is W=3, i.e., each a line picture for closing on edge including edge line He its both sides
Vegetarian refreshments；
312) while increasing width W1, W2 of edge line both sides, each increased step-length is 1, as long as the region of both sides is equal
It is that smooth region width then increases again, is further added by meeting while width W less than or equal to 15；
32) for the image block chosen, 15 are less than or equal to more than or equal to 9 width is met, the gray scale of edge line both sides
When average difference is more than or equal to 66, the image block is selected as the image block calculated for MTF, otherwise given up, be then back to turn
31), until the point in edge image is traveled through completely, then traversal is exited.
In further embodiment, abovementioned steps 2-1) in rim detection, the position for obtaining marginal point is to obtain edge
The sub-pix point position of point, its realization includes：
41) each row to image block carries out simple differencing computing, chooses the maximum of points or minimum of every a line after difference
Value point, i.e. flex point, as the pixel edge point of a line；
42) totally 4 pixels calculate the position of sub-pix point for selected pixels level marginal point and the right and left, and calculating is adopted
This 4 data are fitted with the method for cubic polynomial curve, 0 position of matched curve is the sub-pix side of the row
Edge point position.
In further embodiment, abovementioned steps 2-2) in, include clicking through sub-pix with least square fitting edge
Row fitting a straight line, all marginal points for will being extracted in abovementioned steps are enforceable to be moved on on straight line, so that each
Capable center is marginal point.
In further embodiment, abovementioned steps 2-3) in, the step 6) in the edge in the image block that is fitted is entered
Row interpolation and the spread function treatment of extraction average edge, wherein：
Interpolation processing includes carrying out cubic spline interpolation to each line number strong point of image block, the interpolated resolution for using for
0.05, i.e., 20 interpolation points are inserted between each two data point, so that every a line of image turns into an approximate continuous
Line, its grey value profile is exactly the edge spread function of the row；
Extracting average edge spread function includes the edge spread function of all rows adds up and its average value is taken, and is put down
Equal edge spread function.
In further embodiment, abovementioned steps 2-4) in, simple differencing is done to average spread function, obtain line extension letter
Number, and is fitted with Gaussian Profile to line spread function, its implement including：
In order that the influence of the NF to MTF result of calculations of the smooth region of both sides of edges is tried one's best reduction, first
Region to both sides of edges carries out appropriate interception, that is, intercept each 100 data values of both sides of edges, and letter is extended to the line after interception
Number is fitted using Gaussian Profile to it.
As long as it should be appreciated that all combinations of aforementioned concepts and the extra design for describing in greater detail below are at this
A part for the subject matter of the disclosure is can be viewed as in the case that the design of sample is not conflicting.In addition, required guarantor
All combinations of the theme of shield are considered as a part for the subject matter of the disclosure.
Can be more fully appreciated with from the following description with reference to accompanying drawing present invention teach that foregoing and other aspect, reality
Apply example and feature.The feature and/or beneficial effect of other additional aspects such as illustrative embodiments of the invention will be below
Description in it is obvious, or by according to present invention teach that specific embodiment practice in learn.
Brief description of the drawings
Accompanying drawing is not intended to drawn to scale.In the accompanying drawings, identical or approximately uniform group of each for showing in each figure
Can be indicated by the same numeral into part.For clarity, in each figure, not each part is labeled.
Now, by example and the embodiment of various aspects of the invention will be described in reference to the drawings, wherein：
Fig. 1-4,5-8 exemplarily illustrates 4 remote sensing images and its edge dot image schematic diagram of angle, wherein Fig. 1,
Fig. 2, Fig. 3, Fig. 4 are respectively to carry out 0 degree, 45 degree, 90 degree and 135 degree to original image to rotate the remote sensing images that obtain, Fig. 5, Fig. 6,
Fig. 7, Fig. 8 are respectively that the extraction for carrying out marginal point to Fig. 1, Fig. 2, Fig. 3, Fig. 4 respectively using Canny operators obtains four angles
Edge dot image.
Fig. 9 is that image block chooses flow chart.
Figure 10-15 is image block edge line and fitting result schematic diagram, and Figure 10 is image block and its corresponding gray value
Schematic diagram, Figure 11 is a line gray value schematic diagram in image block, and Figure 12 is the differentiated knot of a line gray value in image block
Fruit schematic diagram, Figure 13 is the intensity profile schematic diagram of single row of pixels, and the difference profile figure of Figure 14 single row of pixels, Figure 15 is examined for edge
The result schematic diagram of survey.
Figure 16 is the schematic diagram of image block single file interpolation.
Figure 17-18 is average edge spread function (ESF) acquisition process schematic diagram, and Figure 17 is all rows (5 in image block
ESF curve synoptic diagrams OK), Figure 18 is the average ESF curve synoptic diagrams with sampled point.
Figure 19-21 is that line spread function (LSF) is obtained and fit procedure schematic diagram, and Figure 19 is to calculate line spread function
(LSF) curve synoptic diagram, Figure 20 is the LSF curve synoptic diagrams after interception, and Figure 21 is that Gaussian Profile is fitted knot to the LSF after interception
Fruit schematic diagram.
Figure 22 is MTF sequence chart schematic diagrames.
Figure 23 is the mtf value schematic diagram that all image blocks are calculated.
Figure 24 is the exemplary process diagram of the multi-angle braking MTF methods of estimation of remote sensing images proposed by the present invention.
Specific embodiment
In order to know more about technology contents of the invention, especially exemplified by specific embodiment and institute's accompanying drawings are coordinated to be described as follows.
Each side with reference to the accompanying drawings to describe the present invention in the disclosure, shown in the drawings of the embodiment of many explanations.
Embodiment of the disclosure must not be intended to include all aspects of the invention.It should be appreciated that various designs presented hereinbefore and reality
Apply example, and those designs for describing in more detail below and implementation method can in many ways in any one come real
Apply, this is to should be design disclosed in this invention to be not limited to any implementation method with embodiment.In addition, disclosed by the invention one
A little aspects can be used alone, or otherwise any appropriately combined be used with disclosed by the invention.
Flow with reference to shown in Figure 24, the automatic MTF methods of estimation of multi-angle of remote sensing images proposed by the present invention, its realization
Chosen including image block and estimate MTF two parts based on recognition status, wherein：
Foregoing image block is chosen and is comprised the following steps：
Original image 1-1) is carried out into multi-angle rotary, postrotational image is obtained；
Marginal point extraction 1-2) is carried out respectively to postrotational image and original image, 4 breadths edge dot images are obtained；
1-3) it is utilized respectively 4 width edge images to refer to, chooses the image block for estimating MTF；
It is foregoing to be comprised the following steps based on recognition status estimation MTF：
2-1) for the image block for having obtained, make rim detection, obtain the position of marginal point, that is, obtain the sub- picture of marginal point
Vegetarian refreshments position；
2-2) with the edge in least square fitting image block；
2-3) enter row interpolation to the edge in the image block of fitting and extract average edge spread function；
Simple differencing 2-4) is done to average spread function, line spread function is obtained, and with Gaussian Profile to line spread function
It is fitted；
Discrete Fourier transform 2-5) is carried out to the line spread function after fitting, and modulus is carried out to result, obtain final
MTF sequences.
With reference to Fig. 1-Figure 23, and the automatic MTF methods of estimation of multi-angle of the remote sensing images shown in Figure 24 realize stream
Journey is illustrated, and illustrates the exemplary implementation of the method.
For remote sensing images to be assessed, original image is carried out into 0 degree, 45 degree, 90 degree and 135 degree 3 rotations of angle first
Turn, so as to obtain the remote sensing images of other 3 angles, as shown in Figure 1, Figure 2, Figure 3, Figure 4.
Four are obtained to the extraction that original image and other 3 images carry out marginal point respectively using Canny (Tuscany) operator
The edge dot image of angle, as shown in Fig. 5, Fig. 6, Fig. 7, Fig. 8.
Next the image block that can be used for MTF calculating in the angle is chosen using the edge image of each angle.For every
One width edge image, each pixel in detection image, if the pixel is marginal point, starts the selection work of image block
Make, otherwise judge next point.The selection of image block is carried out not on edge image, but in corresponding remote sensing images
On.
In this example, abovementioned steps 1-3) the middle image block for utilizing 4 width edge images work to be used to estimate MTF with reference to selection, its
Realization includes：
31) image block of each angle is chosen into the process that process extends as an image block width, for edge graph
Each point of picture, if the point is marginal point, starts the selection process of image block, including:
311) the height H for setting image block is 5, width w self adaptations, and width is by edge line width and edge line both sides
Peak width determines that edge line width is 1, the peak width of the edge line left and right sides its initial value W1=W2=1, both sides relation
It is W=W1+W2+1, initial image block width is W=3, i.e., each a line picture for closing on edge including edge line He its both sides
Vegetarian refreshments；
312) while increasing width W1, W2 of edge line both sides, each increased step-length is 1, as long as the region of both sides is equal
It is that smooth region width then increases again, is further added by meeting while width W less than or equal to 15；
32) for the image block chosen, 15 are less than or equal to more than or equal to 9 width is met, the gray scale of edge line both sides
When average difference is more than or equal to 66, the image block is selected as the image block calculated for MTF, otherwise given up, be then back to turn
31), until the point in edge image is traveled through completely, then traversal is exited.
With reference to shown in Fig. 9, the flow of image block selection is exemplarily given, its realization includes：
Step 1) the height H=5 of image block is set first, the width of the edge line left and right sides is respectively W1=W2=1, and
The width of image block is W=3；
Step 2) judge whether the region of edge line both sides is smooth region, W1 and W2 is increased if being smooth region
1, W increases 2 and turns 3), otherwise to turn 4) accordingly；
Step 3) whether W is judged less than or equal to 15, if meet condition to turn 2), otherwise to turn 4)；
Step 4) W1 and W2 subtracts 1, and whether W accordingly subtracts 2, judges W more than or equal to 9, and the image block is expired if condition is met
Sufficient MTF design conditions, are usable image block, otherwise give up the image block.
By the image block selecting step of the above, the image block of MTF calculating is can be used on 4 directions will be selected,
To be adopted for each width image block carries out the evaluation work of MTF in the following method.
Every a line of image block has a marginal point, and these marginal points constitute the edge line of image block, first step work
Work is exactly the position for obtaining marginal point.
In this example, abovementioned steps 2-1) in rim detection, the position for obtaining marginal point is the sub-pix for obtaining marginal point
Point position, its realization includes：
41) each row to image block carries out simple differencing computing, chooses the maximum of points or minimum of every a line after difference
Value point, i.e. flex point, as the pixel edge point of a line；
42) totally 4 pixels calculate the position of sub-pix point for selected pixels level marginal point and the right and left, and calculating is adopted
This 4 data are fitted with the method for cubic polynomial curve, 0 position of matched curve is the sub-pix side of the row
Edge point position.
It is as shown in fig. 10-15 the process of acquisition data line marginal point, Figure 10 is an image block corresponding with its
Gray value, the height and width of the image block are respectively 5 and 15.If Figure 11 is the data of a line gray value and the distribution situation of gray scale
(with 0~255 span), for the determination of every a line edge point position, as preferred scheme in this example, its realization includes：
Step 1) to data block in every a line gray value do simple differencing, that is, calculate x [n]-x [n-1], such as Figure 11 for figure
As a line gray value in block, Figure 12 is differentiated result, and takes max (x [n]-x [n-1)], it is possible to detect it is maximum tiltedly
The position of rate point, the intensity profile and difference profile figure of the respectively single row of pixels of such as Figure 13,14；
In theory, above-mentioned greatest gradient point, i.e. flex point position are the position of marginal point, but, this result
It is pixel level position, requirement can not be met.In order to further determine that the accurate location of the sub-pixel of the row marginal point, carry out
Step 2).
Step 2) using 4 values of point near flex point, 4 data points of overstriking shown in Figure 11 do cubic polynomial bent
Line is fitted, and fitting formula is as follows：
A (x)=a1+a2x2+a3x+a4
Because second dervative crosses at 0 point, i.e.,：
A " (x)=6a1x+2a2=0
Solve：
X=- (2a2)/(6a1)
X is the marginal point sub-pixel position of the required row, to Figure 10 in every a line all carry out as above step
Treatment, it is possible to obtain the position of the sub-pixel of the marginal point of every a line.The result of rim detection is shown in Figure 15, it is red
The position being by the marginal point sub-pixel detected per one-row pixels of small circle mark, in Figure 15 tables of data brackets
Be often capable sub-pixel edge point position.
Behind the position for obtaining marginal point, further work is exactly the fitting for carrying out edge.One of blade method is substantially false
If being, the edge of blade is located on straight line.Deviation relative to any point point of the straight line all can be to final calculating
MTF results produce influence.So, based on this consideration, all marginal points that will be extracted above are all enforceable straight to one
Come on line, therefore, it is necessary to carry out linear fit to edge detection results.It is least square method to be fitted the method for using.
In this example, abovementioned steps 2-2) in, include carrying out fitting a straight line to sub-pix point with least square fitting edge,
All marginal points for being extracted in abovementioned steps are enforceable to be moved on on straight line, so that the center per a line
It is marginal point.
Assuming that linear equation：
Y=ax+b
Wherein
Wherein m is marginal point number, xiIt is the height of image block, yiIt is sub-pixel edge point position.At this moment to every a line
The relative position of pixel is adjusted, and the standard of adjustment is so that the equal energy in position of each row sub-pixel edge point of first step detection
Accurately fall on fitted straight lines of edges so that each row of data is extended to both sides centered on marginal point.The mesh of do so
, the average of each row edge spread function is conveniently asked, reduce the influence of error.
After being adjusted to each row data, next step is exactly to carry out cubic spline to each line number strong point of image to insert
Value.The interpolated resolution for using is 0.05, i.e., 20 interpolation points are inserted between each two data point.So, image is each
Row just turns into a line for approximate continuous, and its intensity profile is exactly the edge spread function of the row.Figure 16 is shown to single file number
According to the result for carrying out cubic spline interpolation.Because every a line of image all enters row interpolation, therefore, it can draw one per a line
The nearly continuous ESF curves of bar, the ESF of all rows is added up and its average value is taken, it is possible to average edge spread function is obtained
ESF.Figure 17 show the ESF of all rows (5 row) in image block, and Figure 18 is the average ESF curves with sampled point.
In this example, abovementioned steps 2-3) in row interpolation entered to the edge in the image block that is fitted and extract average edge extension
Function treatment, wherein：
Interpolation processing includes carrying out cubic spline interpolation to each line number strong point of image block, the interpolated resolution for using for
0.05, i.e., 20 interpolation points are inserted between each two data point, so that every a line of image turns into an approximate continuous
Line, its grey value profile is exactly the edge spread function of the row；
Extracting average edge spread function includes the edge spread function of all rows adds up and its average value is taken, and is put down
Equal edge spread function.
Once having obtained average edge spread function (ESF), next step is exactly to calculate line spread function (LSF), and it is right only to need
Average ESF does simple differencing, and difference formula is as follows, acquired results such as Figure 19：
LSF (n)=ESF (n)-ESF (n-1)
The image block requirement blade both sides for calculating MTF do not have the interference of too many noise or other target objects, i.e. knife
The intensity profile of the gray scale smooth region of sword both sides of edges is relatively uniform.In order that obtaining the smooth area in knife edge both sides
Influence dosage of the NF in domain to MTF result of calculations is reduced, and the region to both sides of edges carries out appropriate interception here, i.e.,
Interception each 100 data values of both sides of edges, leave behind the crest of edge and the region that both sides are appropriate.Figure 20 is after intercepting
LSF curves.Then Gaussian Profile is fitted to the LSF after interception in, and the fitting result for obtaining is as shown in figure 21.
In this example, abovementioned steps 2-4) in, simple differencing is done to average spread function, obtain line spread function, and with height
This distribution is fitted to line spread function, its implement including：
In order that the influence of the NF to MTF result of calculations of the smooth region of both sides of edges is tried one's best reduction, first
Region to both sides of edges carries out appropriate interception, that is, intercept each 100 data values of both sides of edges, and letter is extended to the line after interception
Number is fitted using Gaussian Profile to it.
After line spread function LSF after being fitted, DFT is carried out to it, take each component after converting
Mould, and on the basis of the DC component after conversion, i.e. first mtf value, normalized is done, just obtain desired MTF sequences
Row.Formula (1) is represented carries out Fourier transform to line spread function, and formula (2) represents the direct current to the result of Fourier transform
Component, i.e. real part do normalized.What formula (3) was represented is to give Nyquist frequency computing formula, Nyquist frequencies
It is the half of cut-off frequency.
MTF (n)=| DFT (LSF (n)) | (1)
Norm_MTF (n)=MTF (n)/MTF (1) (2)
Nyquist_frequency=(whole_data_size × resolution)/2+1
=(Number_of_trimmed_pixel)/2+1 (3)
Wherein DFT () represents DFT, and norm_MTF (n) is normalization mtf value, Nyquist_
Frequency is Nyquist Frequency points, and whole_data_size is that LSF sub-pixels are wide after the interception comprising all interpolation points
Degree, resolution is interpolated resolution, and Number_of_trimmed_pixel is the LSF pixel level widths after interception.Figure 22
The MTF of final calculating is shown, the mtf value result of calculation at Nyquist frequencies is 0.5301.
Figure 23 is the mtf value that all image blocks (35 image blocks) are calculated, for each image block using above-mentioned
Method calculate MTF.
Although the present invention is disclosed above with preferred embodiment, so it is not limited to the present invention.Skill belonging to of the invention
Has usually intellectual in art field, without departing from the spirit and scope of the present invention, when can be used for a variety of modifications and variations.Cause
This, protection scope of the present invention ought be defined depending on those as defined in claim.
Claims (7)
1. a kind of automatic MTF methods of estimation of multi-angle of remote sensing images, it is characterised in that the automatic MTF methods of estimation of the multi-angle
Chosen including image block and estimate MTF two parts based on recognition status, wherein：
Foregoing image block is chosen and is comprised the following steps：
Original image 1-1) is carried out into multi-angle rotary, postrotational image is obtained, its anglec of rotation is respectively：45 degree, 90 degree and
135 degree；
Marginal point extraction 1-2) is carried out respectively to postrotational image and original image, 4 width edge images are obtained；
1-3) it is utilized respectively 4 width edge images to refer to, chooses the image block for estimating MTF；
It is foregoing to be comprised the following steps based on recognition status estimation MTF：
2-1) for the image block for having obtained, make rim detection, obtain the position of marginal point, that is, obtain the sub-pix point of marginal point
Position；
2-2) with the edge in least square fitting image block；
2-3) enter row interpolation to the edge in the image block of fitting and extract average edge spread function；
Simple differencing 2-4) is done to average spread function, line spread function is obtained, and line spread function is carried out with Gaussian Profile
Fitting；
Discrete Fourier transform 2-5) is carried out to the line spread function after fitting, and modulus is carried out to result, obtain final MTF
Sequence.
2. automatic MTF methods of estimation of the multi-angle of remote sensing images according to claim 1, it is characterised in that abovementioned steps
Marginal point extraction is carried out in 1-2), Canny is chosen and is become the extraction for bringing and carrying out marginal point, obtain the edge image of entire image.
3. automatic MTF methods of estimation of the multi-angle of remote sensing images according to claim 1, it is characterised in that abovementioned steps
4 width edge images are utilized to make with reference to the image block chosen for estimating MTF in 1-3), its realization includes：
31) image block of each angle is chosen into the process that process extends as an image block width, for edge image
Each point, if the point is marginal point, starts the selection process of image block, including:
311) the height H for setting image block is 5, width W self adaptations, and width is by edge line width and the region of edge line both sides
Width determines that edge line width is 1, the peak width of the edge line left and right sides its initial value W1=W2=1, and both sides relation is W
=W1+W2+1, initial image block width is W=3, i.e., each one-row pixels for closing on edge including edge line He its both sides
Point；
312) while increasing width W1, W2 of edge line both sides, each increased step-length is 1, as long as the region of both sides is flat
Sliding peak width then increases again, is further added by meeting while width W less than or equal to 15；
32) for the image block chosen, 15 are less than or equal to more than or equal to 9 width is met, the gray average of edge line both sides
When difference is more than or equal to 66, the image block is selected as the image block calculated for MTF, otherwise given up, be then back to turn 31),
Until the point in edge image is traveled through completely, then traversal is exited.
4. automatic MTF methods of estimation of the multi-angle of remote sensing images according to claim 1, it is characterised in that abovementioned steps
Rim detection in 2-1), the position for obtaining marginal point is the sub-pix point position for obtaining marginal point, and its realization includes：
41) each row to image block carries out simple differencing computing, chooses the maximum of points or minimum point of every a line after difference,
That is flex point, as the pixel edge point of the row；
42) totally 4 pixels calculate the position of sub-pix point for selected pixels level marginal point and the right and left, and calculating uses three
The method of order polynomial curve is fitted to this 4 data, and 0 position of matched curve is the sub-pixel edge point of the row
Position.
5. automatic MTF methods of estimation of the multi-angle of remote sensing images according to claim 4, it is characterised in that abovementioned steps
In 2-2), include carrying out fitting a straight line to sub-pix point with least square fitting edge, it is all by what is extracted in abovementioned steps
Marginal point is enforceable to be moved on on straight line, so that the center per a line is marginal point.
6. automatic MTF methods of estimation of the multi-angle of remote sensing images according to claim 5, it is characterised in that abovementioned steps
Enter row interpolation and the spread function treatment of extraction average edge in 2-3) to the edge in the image block of fitting, wherein：
Interpolation processing includes carrying out cubic spline interpolation to each line number strong point of image block, the interpolated resolution for using for
0.05, i.e., 20 interpolation points are inserted between each two data point, so that every a line of image turns into an approximate continuous
Line, its grey value profile is exactly the edge spread function of the row；
Extracting average edge spread function includes the edge spread function of all rows adds up and its average value is taken, and obtains average side
Edge spread function.
7. automatic MTF methods of estimation of the multi-angle of remote sensing images according to claim 6, it is characterised in that abovementioned steps
In 2-4), simple differencing is done to average spread function, obtain line spread function, and line spread function is intended with Gaussian Profile
Close, its implement including：
In order that the influence of the NF to MTF result of calculations of the smooth region of both sides of edges is tried one's best reduction, first opposite side
The region of edge both sides carries out appropriate interception, that is, intercept each 100 data values of both sides of edges, to the line spread function profit after interception
It is fitted with Gaussian Profile.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410769046.XA CN104537646B (en) | 2014-12-12 | 2014-12-12 | The automatic MTF methods of estimation of multi-angle of remote sensing images |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410769046.XA CN104537646B (en) | 2014-12-12 | 2014-12-12 | The automatic MTF methods of estimation of multi-angle of remote sensing images |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104537646A CN104537646A (en) | 2015-04-22 |
| CN104537646B true CN104537646B (en) | 2017-06-27 |
Family
ID=52853165
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410769046.XA Expired - Fee Related CN104537646B (en) | 2014-12-12 | 2014-12-12 | The automatic MTF methods of estimation of multi-angle of remote sensing images |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104537646B (en) |
Families Citing this family (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105069313B (en) * | 2015-08-24 | 2017-11-10 | 北京理工大学 | In-orbit MTF methods of estimation based on phase nonlinear resampling fitting sword side |
| CN105719298B (en) * | 2016-01-22 | 2018-05-29 | 北京航空航天大学 | A kind of method of the line spread function extraction based on edge detecting technology |
| US10067029B2 (en) * | 2016-02-12 | 2018-09-04 | Google Llc | Systems and methods for estimating modulation transfer function in an optical system |
| CN107222683A (en) * | 2017-07-17 | 2017-09-29 | 深圳市东视讯科技有限公司 | Binocular panorama camera produces lens articulation coherence method and system |
| WO2019041590A1 (en) * | 2017-08-31 | 2019-03-07 | 中国科学院微电子研究所 | Edge detection method using arbitrary angle |
| CN108174196B (en) * | 2018-01-15 | 2019-10-18 | 浙江大学 | Based on distance weighted imaging system modulation excitation vibration method |
| CN110807768A (en) * | 2019-10-29 | 2020-02-18 | 核工业北京地质研究院 | Remote sensing image quality evaluation method based on MTF |
| CN112985781B (en) * | 2021-04-27 | 2021-08-10 | 立臻科技(昆山)有限公司 | Lens test data processing method, device, equipment and storage medium |
| CN114022795A (en) * | 2021-11-16 | 2022-02-08 | 北京航空航天大学 | Infrared system MTF automatic calculation method based on real shooting image |
| KR102699108B1 (en) * | 2022-11-17 | 2024-08-26 | 한국항공우주연구원 | Apparatus, method and computer program for measuring modulation transfer function of electro-optical equipment |
| CN116012444B (en) * | 2022-12-05 | 2023-08-18 | 中国科学院长春光学精密机械与物理研究所 | Dynamic image shift compensation bias current curve fitting method |
| CN117953194A (en) * | 2024-01-04 | 2024-04-30 | 中国科学院空天信息创新研究院 | Image enhancement method and device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101281250A (en) * | 2007-04-04 | 2008-10-08 | 南京理工大学 | Method for monitoring on-rail satellite remote sensor modulation transfer function based on image element |
| CN101980293A (en) * | 2010-09-02 | 2011-02-23 | 北京航空航天大学 | Method for detecting MTF of hyperspectral remote sensing system based on edge image |
| CN103679652A (en) * | 2013-11-29 | 2014-03-26 | 北京空间机电研究所 | Image restoration system capable of improving imaging quality greatly |
| CN103970993A (en) * | 2014-04-30 | 2014-08-06 | 中国科学院长春光学精密机械与物理研究所 | Measuring and calculating method for modulation transfer function of satellite-borne camera |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6900884B2 (en) * | 2001-10-04 | 2005-05-31 | Lockheed Martin Corporation | Automatic measurement of the modulation transfer function of an optical system |
-
2014
- 2014-12-12 CN CN201410769046.XA patent/CN104537646B/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101281250A (en) * | 2007-04-04 | 2008-10-08 | 南京理工大学 | Method for monitoring on-rail satellite remote sensor modulation transfer function based on image element |
| CN101980293A (en) * | 2010-09-02 | 2011-02-23 | 北京航空航天大学 | Method for detecting MTF of hyperspectral remote sensing system based on edge image |
| CN103679652A (en) * | 2013-11-29 | 2014-03-26 | 北京空间机电研究所 | Image restoration system capable of improving imaging quality greatly |
| CN103970993A (en) * | 2014-04-30 | 2014-08-06 | 中国科学院长春光学精密机械与物理研究所 | Measuring and calculating method for modulation transfer function of satellite-borne camera |
Non-Patent Citations (5)
| Title |
|---|
| 《An analysis of the knife-edge method for on-orbit MTF estimation of optical sensors》;Xianbin Li等;《International Journal of Remote Sensing》;20100928;第31卷(第17期);全文 * |
| 《Numerical Analysis of MTF for Continuous Scan Imaging Optical System》;REN Zhibin等;《光电工程》;20140930;第41卷(第9期);全文 * |
| 《Switchable electro-optic diffractive lens with high efficiency for ophthalmic applications》;Guoqiang Li等;《PNAS》;20060418;第103卷(第16期);全文 * |
| 《基于刃边法的星载相机在轨MTF测量精度分析》;周川杰等;《航天返回与遥感》;20110228;第32卷(第1期);全文 * |
| 《高分辨光学卫星传感器在轨MTF检测》;徐伟伟等;《大气与环境光学学报》;20140331;第9卷(第2期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104537646A (en) | 2015-04-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104537646B (en) | The automatic MTF methods of estimation of multi-angle of remote sensing images | |
| CN101980293B (en) | Method for detecting MTF of hyperspectral remote sensing system based on edge image | |
| CN104484648B (en) | Robot variable visual angle obstacle detection method based on outline identification | |
| CN101251926B (en) | Remote sensing image registration method based on local configuration covariance matrix | |
| US10613214B2 (en) | Terrestrial imaging using multi-polarization Synthetic Aperture Radar | |
| EP2360642A2 (en) | Video object tracking | |
| CN106197612B (en) | A kind of transparent bottled liquid-level detecting method based on machine vision | |
| CN102609918A (en) | Image characteristic registration based geometrical fine correction method for aviation multispectral remote sensing image | |
| WO2005038394A1 (en) | Refinements to the rational polynomial coefficient camera model | |
| CN101976436A (en) | Pixel-level multi-focus image fusion method based on correction of differential image | |
| CN104899892A (en) | Method for quickly extracting star points from star images | |
| CN102175993A (en) | Radar scene matching feature reference map preparation method based on satellite SAR (synthetic aperture radar) images | |
| CN105139391A (en) | Edge detecting method for traffic image in fog-and-haze weather | |
| CN109741376A (en) | It is a kind of based on improve RANSAC algorithm in, LONG WAVE INFRARED method for registering images | |
| Fergason et al. | Analysis of local slopes at the InSight landing site on Mars | |
| CN108507564B (en) | Star sensor centroid positioning method based on point spread function fitting | |
| CN118260606A (en) | Automatic registration method for ocean remote sensing geographic coordinates of low-orbit satellite | |
| CN113221883A (en) | Real-time correction method for flight navigation route of unmanned aerial vehicle | |
| CN106709941B (en) | A kind of key point screening technique for spectrum image sequence registration | |
| CN105046691A (en) | Method for camera self-calibration based on orthogonal vanishing points | |
| CN105403886B (en) | A kind of carried SAR scaler picture position extraction method | |
| CN111161186B (en) | Push-broom type remote sensor channel registration method and device | |
| US11682142B2 (en) | Information weighted rendering of 3D point set | |
| Pesaresi et al. | Estimating the velocity and direction of moving targets using a single optical VHR satellite sensor image | |
| Crespi et al. | GeoEye-1: Analysis of radiometric and geometric capability |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170627 Termination date: 20211212 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |