CN111932526A - Optical remote sensing load ground resolution automatic detection method based on radial target - Google Patents

Optical remote sensing load ground resolution automatic detection method based on radial target Download PDF

Info

Publication number
CN111932526A
CN111932526A CN202010860235.3A CN202010860235A CN111932526A CN 111932526 A CN111932526 A CN 111932526A CN 202010860235 A CN202010860235 A CN 202010860235A CN 111932526 A CN111932526 A CN 111932526A
Authority
CN
China
Prior art keywords
radius
target
detected
line
row
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202010860235.3A
Other languages
Chinese (zh)
Other versions
CN111932526B (en
Inventor
张静
李晓辉
李传荣
朱家佳
米琳
李子扬
苑馨方
窦帅
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Aerospace Information Research Institute of CAS
Original Assignee
Aerospace Information Research Institute of CAS
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Aerospace Information Research Institute of CAS filed Critical Aerospace Information Research Institute of CAS
Priority to CN202010860235.3A priority Critical patent/CN111932526B/en
Publication of CN111932526A publication Critical patent/CN111932526A/en
Application granted granted Critical
Publication of CN111932526B publication Critical patent/CN111932526B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • 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
    • G06T7/60—Analysis of geometric attributes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/70—Determining position or orientation of objects or cameras
    • G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
    • 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)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Quality & Reliability (AREA)
  • Geometry (AREA)
  • Image Analysis (AREA)

Abstract

An optical remote sensing load external field test ground resolution automatic detection method based on a radial target comprises the following steps: selecting a target sector area to be detected in an image operation interface, wherein the target sector area comprises a target central point O and a radius R of the outermost ring of the area to be detectedmaxAnd RmaxCoordinates P of two end points of corresponding arc segment1、P2The coordinates of (a); based on the region to be detected, extracting image data corresponding to the region to be detected from the image, normalizing the data to 0-255 according to the image quantization bit number to form MfLine by NfA data matrix of columns; acquiring a corresponding row number of the target starting position in the data matrix, marking the row number as WSL, and taking the radius corresponding to the row as a starting radius RWSL(ii) a Just right on target image is obtained by adopting white line detection methodThe corresponding line number of the resolution position in the data matrix is marked as LML, and the radius corresponding to the line is the limit radius RLML(ii) a According to the starting radius RWSLLimit radius RLMLTarget outer ring chord length DWSLAnd calculating the ground resolution according to the target strip angle width theta.

Description

Optical remote sensing load ground resolution automatic detection method based on radial target
Technical Field
The invention relates to the technical field of optical image detection, in particular to an automatic detection method for ground resolution in an optical remote sensing load outfield test based on a radial target.
Background
The ground resolution of the optical remote sensing load refers to the minimum ground distance or the size of a minimum target object which can be resolved on a remote sensing image acquired by the load. The ground resolution of the optical remote sensing load embodies the capability of the camera, the photosensitive film, the CCD sensor or the whole imaging system to realize the on-line object details. The ground resolution is an important index for testing and evaluating the optical remote sensing load performance, the image quality and the application efficiency no matter ground imaging, aerial photography and space remote sensing.
At present, laboratory evaluation and measurement of optical camera and optical remote sensing load resolution are generally carried out by visually interpreting a minimum resolution unit on a resolution target image acquired by the optical camera or the optical remote sensing load to determine the resolution. Commonly used laboratory resolution test targets are described below. FIG. 1(a) is a USFA 1951, a standard test target issued by air force in a country, primarily for laboratory resolution detection of images taken by film cameras; fig. 1(b) shows the test targets given by the ISO12233 standard for resolution assessment in the civil still photograph digital camera laboratory, and the wedge targets are in the gray frame, and have a difference between the 9-line target and the 5-line target, which can be used for measuring resolution detection in three directions, namely horizontal, vertical and 45-degree inclined angles, and fig. 1(d) shows the 5-line wedge targets in the ISO12233 resolution board in fig. 1 (b).
The outfield test optical remote sensing load ground resolution also refers to a laboratory resolution test method, and the minimum distinguishable dimension of an artificial ground object or a natural ground object (such as a road and an automobile) is determined by interpreting images of the artificial ground object (such as a resolution target) or the natural ground object (such as a road and an automobile) with typical structural characteristics, which are obtained by the optical remote sensing load, so that the ground resolution of the optical remote sensing load in operation is evaluated. The ground resolution of the optical remote sensing load is evaluated by utilizing the artificial ground object, a resolution target is generally required to be arranged on the ground, a three-line resolution target or a radial target is typically compared, and the ground resolution of an image is measured by judging the width of the minimum distinguishable line of the target. FIG. 1(c) shows a radial target with NASA deployed in the Stennis space center for out-field testing of ground resolution for in-orbit high resolution optical telemetry loads (IKONOS, Quickbird, SPOT, etc.).
The evaluation of the ground resolution by using the visual interpretation method is relatively intuitive, but the method is a subjective perception process, is easily influenced by human factors (such as emotion, experience, learning and the like of an interpreter) and image output equipment (such as a display, a printer and the like), is time-consuming and tedious, cannot ensure the repeated reproducibility of an interpretation result, and is difficult to systematize and engineer in practical application. Therefore, the practical image objective data-based high-optical remote sensing load outfield test ground resolution automatic detection method which is consistent with visual interpretation and evaluation results is researched and developed, and the method has urgent practical requirements on the aspects of on-orbit running, outfield test and load performance evaluation of the current optical remote sensing load.
Some documents disclose laboratory detection methods of digital camera resolution, and most typically represent a digital camera resolution detection algorithm based on a wedge target (referred to as "CIPA algorithm") in the CIPA DC-003 standard.
The consistency of the CIPA algorithm and the visual interpretation evaluation result is high after verification, but the method is a laboratory evaluation method, is not applied to the field test of the ground resolution, and has the limitations that:
a) the CIPA algorithm is only used for determining the resolution in the horizontal or vertical direction (using the target shown in fig. 1 (d)), and if the resolution in other directions (such as the inclination of 45 °) is to be determined, the image needs to be rotated in advance by adopting a pre-interpolation algorithm, and the pre-interpolation algorithm is not suitable for the optical remote sensing loading outfield test situation. Errors may be introduced if the evaluation is performed by rotating the image.
b) The target used by the algorithm is a white background black line, and the target is usually set to be a black background white line under the condition of the optical remote sensing load external field test to inhibit the influence of the edge effect, so a white line detection method is required, and the detection threshold value is required to be reset.
c) The CIPA algorithm requires that the ratio of the maximum and minimum reflectivity (or gray value) of an image to be detected is greater than 18, the minimum threshold used in line detection is 0 at this time, and for a remote sensing image obtained by an optical remote sensing load, this condition is not always met, and an accurate evaluation result may not be obtained.
At present, radial targets with alternate black and white lines are generally adopted for the ground resolution in the test and evaluation of the optical remote sensing load outfield, and the CIPA method is not suitable, so that an automatic detection method for the ground resolution in the test of the optical remote sensing load outfield based on the radial targets is necessary to be designed.
Disclosure of Invention
In view of the above, the present invention provides an automatic ground resolution detection method for an optical remote sensing loading outfield test based on a radial target, so as to solve at least one of the above technical problems.
In order to achieve the above object, as an aspect of the present invention, there is provided an automatic detection method for ground resolution in an optical remote sensing load external field test based on a radial target, comprising the following steps:
selecting a region to be detected: selecting a target sector area to be detected in an image operation interface, wherein the target sector area comprises a target central point O and a radius R of the outermost ring of the area to be detectedmaxAnd RmaxCoordinates P of two end points of corresponding arc segment1、P2The coordinates of (a); o, P therein1、P2The coordinates of (a) are its row and column number in the image;
extracting data of the area to be detected: based on the region to be detected, extracting image data corresponding to the region to be detected from the image, normalizing to 0-255 according to the image quantization bit number to form MfLine by NfA data matrix of columns;
and (3) detecting an initial line: acquiring a corresponding row number of the target starting position in the data matrix, marking the row number as WSL, and taking the radius corresponding to the row as a starting radius RWSL；
And (3) limit line detection: adopting a white line detection method to obtain a line number corresponding to the exactly resolvable position on the target image in the data matrix, marking the line number as LML, wherein the radius corresponding to the line is the limit radius RLML(ii) a And
and (3) calculating the ground resolution: according to the starting radius RWSLLimit radius RLMLTarget outer ring chord length DWSLAnd calculating the ground resolution according to the target strip angle width theta.
The radial target is a sector area formed by alternately arranging a plurality of white and black sector target strips on a black substrate.
Wherein the limiting radius R of the radial targetLMLIt can also be obtained by averaging a plurality of measured extreme radii, including:
selection of a radial target containing NWLTaking the fan-shaped area of each white target strip as an area to be detected;
the outermost ring radius R corresponding to the outermost ring position of the sector areamaxAs an initial detection radius, taking the leftmost point of the arc segment corresponding to the detection radius as a starting point and the rightmost point as an end point, acquiring the image gray value of each point on the arc segment, and judging the number of white target strips contained in the arc segment according to an empirical threshold;
gradually reducing the detection radius r by taking the delta r as a stepping value, repeating the steps, and when the number of the white target strips detected on the arc section corresponding to the detection radius is just less than NWLWhen the strip is in a state of strip, the detection radius is considered as a limit radius RLML1；
Selecting N-1 additional N-containing targets on the radial targetWLA sector area of a white target strip, heavyRepeating the above steps to obtain RLML2、RLML3、RLML4......RLMLn(ii) a And
for the limiting radius R obtained by the above measurementLML2、RLML3、RLML4......RLMLnAveraging to obtain RLML。
Wherein, the selection of the region to be detected specifically comprises:
marking a central point O of the radial target in an image interface by using an image interaction tool;
marking point P outside the outermost ring position arc of the target1Drawing a line segment OP in the interface1Wherein OP1＝RmaxAnd has Rmax＞RWSL；
At OP1Marking point P in clockwise direction, marking radius OP in interface1The intersection point OP of the arc and the line segment OP2Wherein OP1＝OP2＝RmaxRequires P1、P2The circular arc segment between should contain complete NWLWhite target bar, default NWL＝5；
Radius OP1、OP2And P1P2The arc segment defines a sector OP to be detected1P2。
The extraction of the data of the area to be detected specifically comprises the following steps:
calculating the number of rows and columns of the data matrix: according to sector area OP to be detected1P2Maximum radius R ofmaxMinimum radius RminAnd the radius stepping value delta r is summed, and the total row number M of the data matrix is obtained by calculationf＝1+(Rmax-Rmin) A/Δ R, wherein Rmax、RminThe unit of Δ r is pixel; according to the left edge angle of the detection area
Figure BDA0002646590540000041
I.e. radius OP1Corresponding angle, right edge angle
Figure BDA0002646590540000042
I.e. radius OP2Corresponding angle and angle step value
Figure BDA0002646590540000043
Calculating the total number of columns of the target data matrix
Figure BDA0002646590540000044
Wherein the content of the first and second substances,
Figure BDA0002646590540000045
Figure BDA0002646590540000046
the unit of (a) is an angle;
data matrix row data extraction: a circular arc segment corresponding to the radius R of the target, wherein Rmin≥r≥Rmin. Firstly, according to a nearest principle, namely selecting a pixel closest to the calculated coordinate distance, determining a gray value of a leftmost end point of the arc segment, and filling the gray value into the 1 st row and the 1 st column of the data matrix; then is provided with
Figure BDA0002646590540000047
Determining the gray value of each point of the arc segment to the right according to the nearest principle as an angle stepping value, and filling the gray value into each column of the ith row of the data matrix in sequence; and
and (3) data matrix composition: from i to M in sequencefCarrying out data extraction on each data line of the data matrix in rows to finally form an MfLine by NfA data matrix of columns; each row of the data matrix corresponds to a different target radius.
The initial row detection is realized by traversing each row mean value in the acquired data matrix, and the initial row corresponds to the first row of which the row mean value is greater than the background black level.
Wherein the extreme radius detection comprises:
white line detection; and
selecting a minimum threshold; and/or
The white line detection is carried out by adopting a maximum value identification algorithm, the gray change of each row of data in the data matrix is analyzed line by line, the number of the gray maximum values in the data is detected, and each time one gray maximum value is detected, a white target strip is considered to be detected; and/or
The threshold value used for the white line detection judgment needs to accord with the human eye perception rule, is the lowest limit of visual perception, and needs to be determined according to the lowest distinguishable contrast of human eyes under different background conditions.
Wherein, the calculation formula for calculating the ground resolution is as follows:
Figure BDA0002646590540000051
wherein D isWSLRepresenting the outer ring chord length. According to the Nyquist sampling theory, the minimum distinguishable line width should not be less than 1 ground sampling pixel (GSD), which requires a limiting radius
Figure BDA0002646590540000052
So when R isminWhen the resolution is 0, the calculation formula of the ground resolution of the external field test remote sensing image is shown as the following formula:
Figure BDA0002646590540000053
as another aspect of the present invention, there is provided a computer apparatus including:
one or more processors;
a memory to store one or more instructions that,
wherein the one or more instructions, when executed by the one or more processors, cause the one or more processors to implement the method as described above.
As a further aspect of the invention, there is provided a computer readable storage medium having stored thereon executable instructions which, when executed by a processor, cause the processor to implement the method as described above.
Based on the technical scheme, compared with the prior art, the method for automatically detecting the ground resolution in the optical remote sensing load external field test based on the radial target has at least one or part of the following beneficial effects:
(1) the invention provides a method for automatically measuring and calculating the ground resolution of an optical remote sensing load in an outfield test process by utilizing a radial target, solves the problem that the ground resolution evaluation depends on manual interpretation for a long time, and realizes the rapid and objective evaluation of the index. The noise immunity of the method is enhanced by adopting a white line detection principle, the measurement and calculation precision is improved by using sub-pixel radius stepping, and the evaluation stability is improved by multi-region calculation. Through practice verification for many times, the evaluation result of the method has good consistency with the visual interpretation result, and the consistency can reach 96% through measurement and calculation.
(2) The method can also be used for radial target design, and sensitivity analysis and influence factor analysis are carried out on the calculated target design by setting different parameters by the method, so that optimization of radial target layout is facilitated.
(3) The method can also be applied to rapid automatic detection of the SAR remote sensing imaging load outfield ground resolution.
Drawings
FIG. 1(a) is a three-line target as defined in the national Standard USAF1951, FIG. 1(b) is the ISO12233 Standard resolution target, FIG. 1(c) is a radial target with NASA placed in the Stennis space center, FIG. 1(d) is the 5-line wedge target in the ISO12233 resolution plate of FIG. 1 (b);
FIG. 2 is a schematic illustration of a radial target structure in an embodiment of the present invention;
fig. 3(a) is a schematic diagram of a plurality of sectors to be detected in the embodiment of the present invention, and fig. 3(b) is a schematic diagram of a determination process of a single sector to be detected;
FIG. 4 is a flow chart of a method for automatic detection of a single-sector limiting radius based on a radial target according to an embodiment of the present invention;
FIG. 5 is a flow chart of start line detection in an embodiment of the present invention;
FIG. 6 is a flow chart of limit row detection in an embodiment of the present invention;
FIG. 7 is a flow chart of white line detection according to an embodiment of the present invention;
FIG. 8 is a graph of distinguishable contrast thresholds as a function of background gray scale value in an embodiment of the present invention.
Detailed Description
The invention provides a ground resolution automatic detection method based on a radial target by taking the design idea of a wedge-shaped target resolution automatic detection algorithm in the CIPA DC-003 standard as reference and comprehensively considering the human eye visual threshold value aiming at the structural characteristics of the radial target.
Specifically, the invention discloses an automatic detection method for ground resolution in an optical remote sensing load outfield test based on a radial target, which comprises the following steps:
selecting a region to be detected: selecting a target sector area to be detected in an image operation interface, wherein the target sector area comprises a target central point O and a radius R of the outermost ring of the area to be detectedmaxAnd RmaxCoordinates P of two end points of corresponding arc segment1、P2The coordinates of (a); o, P therein1、P2The coordinates of (a) are its row and column number in the image;
extracting data of the area to be detected: based on the region to be detected, extracting image data corresponding to the region to be detected from the image, normalizing to 0-255 according to the image quantization bit number to form MfLine by NfA data matrix of columns;
and (3) detecting an initial line: acquiring a corresponding row number of the target starting position in the data matrix, marking the row number as WSL, and taking the radius corresponding to the row as a starting radius RWSL；
And (3) limit line detection: adopting a white line detection method to obtain a line number corresponding to the exactly resolvable position on the target image in the data matrix, marking the line number as LML, wherein the radius corresponding to the line is the limit radius RLML(ii) a And
and (3) calculating the ground resolution: according to the starting radius RWSLLimit radius RLMLTarget outer ring chord length DWSLAnd calculating the ground resolution according to the target strip angle width theta.
The radial target is a sector area formed by alternately arranging a plurality of white and black sector target strips on a black substrate.
Wherein the limiting radius R of the radial targetLMLIt can also be obtained by averaging a plurality of measured extreme radii, including:
selection of a radial target containing NWLTaking the fan-shaped area of each white target strip as an area to be detected;
the outermost ring radius R corresponding to the outermost ring position of the sector areamaxAs an initial detection radius, taking the leftmost point of the arc segment corresponding to the detection radius as a starting point and the rightmost point as an end point, acquiring the image gray value of each point on the arc segment, and judging the number of white target strips contained in the arc segment according to an empirical threshold;
gradually reducing the detection radius r by taking the delta r as a stepping value, repeating the steps, and when the number of the white target strips detected on the arc section corresponding to the detection radius is just less than NWLWhen the strip is in a state of strip, the detection radius is considered as a limit radius RLML1；
Selecting N-1 additional N-containing targets on the radial targetWLRepeating the above steps to obtain RLML2、RLML3、RLML4......RLMLn(ii) a And
for the limiting radius R obtained by the above measurementLML2、RLML3、RLML4......RLMLnAveraging to obtain RLML。
Wherein, the selection of the region to be detected specifically comprises:
marking a central point O of the radial target in an image interface by using an image interaction tool;
marking point P outside the outermost ring position arc of the target1Drawing a line segment OP in the interface1Wherein OP1＝RmaxAnd has Rmax＞RWSL；
At OP1Marking point P in clockwise direction, marking radius OP in interface1The intersection point OP of the arc and the line segment OP2Wherein OP1＝OP2＝RmaxRequires P1、P2The circular arc segment between should contain complete NWLWhite target bar, default NWL＝5；
Radius OP1、OP2And P1P2The arc segment defines a sector OP to be detected1P2。
The extraction of the data of the area to be detected specifically comprises the following steps:
calculating the number of rows and columns of the data matrix: according to sector area OP to be detected1P2Maximum radius R ofmaxMinimum radius RminAnd the radius stepping value delta r is summed, and the total row number M of the data matrix is obtained by calculationf＝1+(Rmax-Rmin) A/Δ R, wherein Rmax、RminThe unit of Δ r is pixel; according to the left edge angle of the detection area
Figure BDA0002646590540000081
Radius OP1 corresponding to angle, right edge angle
Figure BDA0002646590540000082
I.e. radius OP2Corresponding angle and angle step value
Figure BDA0002646590540000083
Calculating the total number of columns of the target data matrix
Figure BDA0002646590540000084
Wherein the content of the first and second substances,
Figure BDA0002646590540000085
Figure BDA0002646590540000086
the unit of (a) is an angle;
data matrix row data extraction: a circular arc segment corresponding to the radius R of the target, wherein Rmin≥r≥Rmin. Firstly, according to the most adjacent principle, namely selecting the pixel with the closest distance to the calculated coordinate, determining the leftmost end of the circular arc segmentFilling the gray value of the point into the 1 st row and the 1 st column of the data matrix; then is provided with
Figure BDA0002646590540000087
Determining the gray value of each point of the arc segment to the right according to the nearest principle as an angle stepping value, and filling the gray value into each column of the ith row of the data matrix in sequence; and
and (3) data matrix composition: from i to M in sequencefCarrying out data extraction on each data line of the data matrix in rows to finally form an MfLine by NfA data matrix of columns; each row of the data matrix corresponds to a different target radius.
The initial row detection is realized by traversing each row mean value in the acquired data matrix, and the initial row corresponds to the first row of which the row mean value is greater than the background black level.
Wherein the extreme radius detection comprises:
white line detection; and
selecting a minimum threshold; and/or
The white line detection is carried out by adopting a maximum value identification algorithm, the gray change of each row of data in the data matrix is analyzed line by line, the number of the gray maximum values in the data is detected, and each time one gray maximum value is detected, a white target strip is considered to be detected; and/or
The threshold value used for the white line detection judgment needs to accord with the human eye perception rule, is the lowest limit of visual perception, and needs to be determined according to the lowest distinguishable contrast of human eyes under different background conditions.
Wherein, the calculation formula for calculating the ground resolution is as follows:
Figure BDA0002646590540000091
wherein D isWSLRepresenting the outer ring chord length. According to the Nyquist sampling theory, the minimum distinguishable line width should not be less than 1 ground sampling pixel (GSD), which requires a limiting radius
Figure BDA0002646590540000092
So when R isminWhen the resolution is 0, the calculation formula of the ground resolution of the external field test remote sensing image is shown as the following formula:
Figure BDA0002646590540000093
the invention also discloses a computer device, comprising:
one or more processors;
a memory to store one or more instructions that,
wherein the one or more instructions, when executed by the one or more processors, cause the one or more processors to implement the method as described above.
The invention also discloses a computer readable storage medium having stored thereon executable instructions which, when executed by a processor, cause the processor to implement the method as described above.
In order that the objects, technical solutions and advantages of the present invention will become more apparent, the present invention will be further described in detail with reference to the accompanying drawings in conjunction with the following specific embodiments.
The radial target is formed by arranging a plurality of black or white target strips at intervals to form a fan shape, and the structural schematic diagram of the target strips is shown in FIG. 2. The angular width (called angular width for short) theta of the target strip refers to the central angle of a single target strip, the target radiation angle refers to the central angle covered by the whole target, the width d of the target strip refers to the chord length of the single target strip at the radius r, and d is 2 r sin (theta/2). Since θ is usually small (< 10 °), d ≈ r ≈ θ.
For radial targets, the target swath width varies at different radii of the target, given a known radius r1Has a target strip width of d1Then the target strip width d at any radius r is:
Figure BDA0002646590540000101
by definition, the ground resolution RES is equal to the width of a pair of black and white bars that are just resolved on the image, so for a radial target, the ground resolution RES is equal to the right resolvable position of the target (corresponding to radius R)LML) Width d of black and white target strip 02 times of the total weight of the powder.
RES＝RLML×θ (2)
Wherein the radius RLML(in units of pixels) is called the limiting radius.
In the embodiment of the invention, the automatic detection method for the ground resolution based on the external field test of the radial target comprises the following steps:
1. selection of a radial target containing NWLThe fan-shaped area of each white target bar is used as the area to be detected, as shown in fig. 3 (a).
2. Taking the initial radius of the sector as the detection radius, R ═ RWSLThen, the leftmost point of the arc segment corresponding to the detection radius is taken as a starting point, the rightmost point is taken as an end point, the image gray value of each point on the arc segment is obtained, and the number of the white target strips contained in the arc segment is judged according to an empirical threshold.
3. Gradually reducing the detection radius r by taking the delta r as a step length, repeating the steps, and when the number of the white target strips of the arc section corresponding to the detected detection radius is just less than NWLWhen the strip is in a state of strip, the detection radius is considered as a limit radius RLML1。
4. Selecting 3-4 fan-shaped areas containing 5 white target strips on the target, and repeating the steps to obtain RLML2、RLML3、RLML4、RLML5。
5. Averaging each limiting radius to obtain RLML。
6. The ground resolution RES is calculated according to equation (2).
A flowchart of the single sector area ground resolution automatic detection method is shown in fig. 4, and specifically includes the following steps:
A. selecting a region to be detected: determining parameters of the region to be detected, including determining a target center point O and the outermost ring half of the region to be detectedDiameter Rmax，RmaxCoordinates P of two end points of corresponding arc segment1、P2；
B. Extracting data of the area to be detected: obtaining a sector area OP to be detected in an image1P2Forming a data matrix according to the corresponding target image data;
C. and (3) detecting an initial line: acquiring a corresponding row number of the target starting position in the data matrix acquired in the step B, wherein the corresponding radius of the row is the starting radius RWSL；
D. And (3) limit line detection: b, obtaining a row number corresponding to the position which can be just resolved on the target image in the data matrix obtained in the step B through white line detection, wherein the corresponding radius of the row is the limit radius RLML。
E. And (3) calculating the ground resolution: according to the starting radius RWSLLimit radius RLMLTarget outer ring chord length DWSL(known at target placement), target bar angular width θ calculates ground resolution.
Table 1 shows the representative meanings of the parameters and the related calculation formulas used in the algorithm.
TABLE 1 description of parameters in the automated inspection Algorithm
Figure BDA0002646590540000111
Figure BDA0002646590540000121
The method for automatically detecting the ground resolution in the outfield test based on the radial target comprises the following specific implementation steps.
Step A: selection of regions to be detected
The process of selecting the region to be detected is shown in fig. 3(b), and the specific steps are as follows:
marking a central point O of the radial target in an image interface by using an image interaction tool;
at the starting position of the target (corresponding to radius R)WSL) Outer mark point P of arc1The radius OP is plotted in the figure1(OP1＝RmaxAnd has Rmax＞RWSL)；
At OP1Is marked with a point P in the clockwise direction, and a radius OP is marked in the figure1The intersection point OP of the arc and the radius OP2Requires P1、P2The circular arc segment between should contain complete NWLWhite target bar, default NWL＝5。
Radius OP1、OP2And P1P2The arc segment defines a sector OP to be detected1P2。
And B: region data extraction to be detected
Sector area OP to be detected determined according to step A1P2And extracting corresponding data from the image to form a data matrix. Considering that the resolution of the radial target corresponds to the limit radius, when the resolution detection is performed, the rows of the data matrix take the radius as a defining unit, the columns take the angular position as a defining unit, and the stored data are gray values acquired from the image.
Calculating the number of rows and columns of the data matrix: according to sector area OP to be detected1P2Maximum radius R ofmaxMinimum radius Rmin(ideally the radial target is a sector, R min0, but some radial targets cannot approach the center of a circle infinitely due to manufacturing process limitations, and only one fan-shaped ring is formed, namely RminNot equal to 0) and a radius step value Δ r, the total number of rows M of the data matrix is calculatedf＝1+(Rmax-Rmin) A/Δ R, wherein Rmax、RminΔ r is in pixels, and Δ r is usually set to 0.1 pixel; according to the left edge angle of the detection area
Figure BDA0002646590540000131
(radius OP)1Corresponding angle), right edge angle
Figure BDA0002646590540000132
(radius OP)2Corresponding angle), angleStep value
Figure BDA0002646590540000133
Calculating the total number of columns of the target data matrix
Figure BDA0002646590540000134
Wherein the content of the first and second substances,
Figure BDA0002646590540000135
the unit of (a) is an angle,
Figure BDA0002646590540000136
typically set at 0.5 °;
calculating the radius r of the corresponding target in each row of the data matrix: from sector OP to be detected1P2Maximum radius R ofmaxStarting with line 1 of the data matrix, stepping by Δ R and gradually decreasing to the minimum radius RminCorresponding to row number MfThe target radius R corresponding to the ith row of the data matrix is Rmax-Δr*(i-1)。
Data matrix row data extraction: along the arc segment corresponding to the radius r of the target, firstly, determining the gray value of the leftmost end point according to the nearest principle (namely, selecting the pixel closest to the calculated coordinate distance), and filling the gray value into the 1 st row and the 1 st column of the data matrix; then is provided with
Figure BDA0002646590540000137
Determining the gray value of each point of the arc segment according to the nearest neighbor principle for the angle stepping value, and filling the gray value into each column of the ith row of the data matrix in sequence.
And (3) data matrix composition: from i to M in sequencefExtracting data of each data line of the data matrix in rows, normalizing the data to 0-255 according to the quantization bit number of the image, and finally forming an MfLine by NfA data matrix of columns. Each row of the data matrix corresponds to a different target radius.
And C: WSL detection of the starting line (where the starting radius is located)
The step is to automatically obtain the distance from the initial position of the radial target to the circle center, and is calledStarting radius, in RWSLAnd (4) showing. Is provided with
Figure BDA0002646590540000138
The WSL detection flow for the start line is shown in fig. 5. Traversing each row from the first row of the data matrix, setting a detection threshold (background black level) BBL to be 1.5 times of the mean value of the row when detecting that the mean value of the corresponding row is more than 20, and continuously traversing each row until marking the initial row WSL of the behavior target when detecting that the mean value of the corresponding row is more than the detection threshold BBL.
Step D: limit (limit radius) LML detection
The step is to obtain the distance from the exactly resolvable position to the center of the circle on the radial target, called the limiting radius, denoted as RLMLAnd (4) showing. Is provided with
Figure BDA0002646590540000141
The limit line LML detection flow is shown in fig. 6, and mainly includes 3 sub-steps of white line detection, minimum threshold selection, and limit line marking.
Substep D01: white line detection
The white line detection uses two benchmarks at the time of visual assessment:
a. number of white lines N for detecting existence of one objectWLWhen the number of detection lines of the visual resolution evaluation pattern changes (from N)WLBecomes less than NWL) The corresponding position of time is taken as the extreme radius.
b. When observing, the tracking starts from the low-frequency side to the high-frequency side, namely the tracking from the outside to the inside.
The main flowchart of the white line detection step is shown in fig. 7. Wherein, the number of the target white line detection pieces NWLIs a value, N, which needs to be manually set in advance when white line detection is performedWLDetermines the sector area OP to be detected1P2Is also an important parameter for determining the limit radius.
The white line detection method is carried out by adopting a maximum value identification algorithm, the gray change of each line of data in the data matrix is analyzed line by line, the number of the gray maximum values in the data is detected, and each time one gray maximum value is detected, a white target strip is considered to be detected (step S101 in figure 7, the change from 0 to 1 is used as a mark for detecting the white line). The algorithm also considers the suppression of noise in the data and detects only an increase exceeding a threshold value from a local minimum value or a decrease exceeding a set threshold value from a local maximum value as an effective increase or decrease. The local maximum value LMx and the local minimum value LMn are referred to as "maximum value and minimum value of data up to the present since the last detection of an effective change", because when the data value is reset when the change is detected, but no effective change is detected, the data value is updated to the value when the data value is smaller than LMn and the LMn is updated to the value when the data value is larger than LMx.
Threshold value T for white line detection determinationETHThe initial value is set to 1/4 of the maximum value of the start line (WSL) data. Then TETHThe threshold value is updated in the main flow of the program and is gradually and properly reduced according to the requirement. When black and white detection of low-frequency data lines is carried out in the initial stage, the threshold value T is setETHThe correct result can be detected by setting the threshold value to a relatively large value, and the threshold value T is used when the detection range gradually moves to a high-frequency data lineETHThe specified number of lines (N) cannot be detectedWL) Will decrease the threshold value TETHAnd the detection is repeated. When T isETHReduced to a minimum threshold TBIn this case, N is still not detectedWLAnd when the target bar is white, the data line is considered to correspond to a limit line (LML). Minimum threshold TBThe requirement of conforming to the human eye perception rule is the lowest limit of the visual perception, and the requirement is determined according to the lowest distinguishable contrast of human eyes under different background conditions.
Substep D02: minimum threshold TBSelecting
The perception of image information by the human eye needs to be under appropriate contrast conditions. The contrast in the present invention is defined as an absolute value of a gray difference between a background and a target. The object cannot be perceived by the human eye, mainly because the contrast between the object and the background is not achievedTo a human eye distinguishable contrast threshold. The eye-discernable contrast threshold is a subjective perception quantity that characterizes the critical value of the gray-scale change or gray-scale difference that the human eye can just perceive. Minimum threshold T of the inventionBThe research result in the document "human visual contrast resolution limit determination based on digital image processing" is cited, and the minimum threshold value T applied to the invention is appropriately modifiedBAnd (4) calculating as shown in formula (4).
Figure BDA0002646590540000151
Wherein X represents the minimum gray value of a certain row of data in the data matrix, TBRepresenting the resolvable contrast threshold when the background is X. FIG. 8 shows a distinguishable contrast threshold TBGraph with X.
Step E: ground resolution calculation
Using the starting radius RWSLLimit radius RLMLAnd the outer ring chord length D of the target is laidWSL(known at deployment) the ground resolution is calculated according to equation (5).
Figure BDA0002646590540000152
When Rmin is 0, the calculation formula of the resolution can be simplified as follows:
Figure BDA0002646590540000153
according to the Nyquist sampling theory, the minimum distinguishable line width should not be less than 1 ground sampling pixel (GSD), which requires a limiting radius
Figure BDA0002646590540000154
Therefore, the calculation formula of the out-field test optical remote sensing load ground resolution is shown as the following formula:
Figure BDA0002646590540000161
furthermore, the above definitions of the methods are not limited to the various specific structures, shapes or arrangements mentioned in the examples, which may be easily modified or substituted by a person of ordinary skill in the art, for example:
(1) the number N of the target white line detection in the step A and the step DWLThe number of the selected areas can be adjusted to 4 or 6, and the selected areas in the step A also need to be changed.
(2) The radius step value and the angle step value in step B may be changed.
(3) The minimum threshold selection method in step D02 may be replaced with other human-eye recognizable contrast threshold models.
The above-mentioned embodiments are intended to illustrate the objects, technical solutions and advantages of the present invention in further detail, and it should be understood that the above-mentioned embodiments are only exemplary embodiments of the present invention and are not intended to limit the present invention, and any modifications, equivalents, improvements and the like made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (10)

1. An optical remote sensing load external field test ground resolution automatic detection method based on a radial target comprises the following steps:
selecting a region to be detected: selecting a target sector area to be detected in an image operation interface, wherein the target sector area comprises a target central point O and a radius R of the outermost ring of the area to be detectedmaxAnd RmaxCoordinates P of two end points of corresponding arc segment1、P2The coordinates of (a); o, P therein1、P2The coordinates of (a) are its row and column number in the image;
extracting data of the area to be detected: based on the region to be detected, extracting image data corresponding to the region to be detected from the image, normalizing the data to 0-255 according to the image quantization bit number to form MfLine by MfData moments of columnsArraying;
and (3) detecting an initial line: acquiring a corresponding row number of the target starting position in the data matrix, marking the row number as WSL, and taking the radius corresponding to the row as a starting radius RWSL；
And (3) limit line detection: adopting a white line detection method to obtain a line number corresponding to the exactly resolvable position on the target image in the data matrix, marking the line number as LML, wherein the radius corresponding to the line is the limit radius RLML(ii) a And
and (3) calculating the ground resolution: according to the starting radius RWSLLimit radius RLMLTarget outer ring chord length DWSLAnd calculating the ground resolution according to the target strip angle width theta.
2. The automated inspection method of claim 1, wherein said radial targets are sectors of alternating white and black sectors of target strips on a black substrate.
3. The automated method of claim 1, wherein the radial target has a limiting radius RLMLIt can also be obtained by averaging a plurality of measured extreme radii, including:
selection of a radial target containing NWLTaking the fan-shaped area of each white target strip as an area to be detected;
the outermost ring radius R corresponding to the outermost ring position of the sector areamaxAs an initial detection radius, taking the leftmost point of the arc segment corresponding to the detection radius as a starting point and the rightmost point as an end point, acquiring the image gray value of each point on the arc segment, and judging the number of white target strips contained in the arc segment according to an empirical threshold;
gradually reducing the detection radius r by taking the delta r as a stepping value, repeating the steps, and when the number of the white target strips detected on the arc section corresponding to the detection radius is just less than NWLWhen the strip is in a state of strip, the detection radius is considered as a limit radius RLML1；
Selecting N-1 additional N-containing targets on the radial targetWLRepeating the above steps to obtain RLML2、RLML3、RLML4......RLMLn(ii) a And
for the limiting radius R obtained by the above measurementLML2、RLML3、RLML4......RLMLnAveraging to obtain RLML。
4. The automatic detection method according to claim 1, wherein the selection of the region to be detected specifically comprises:
marking a central point O of the radial target in an image interface by using an image interaction tool;
marking point P outside the outermost ring position arc of the target1Drawing a line segment OP in the interface1Wherein OP1＝RmaxAnd has Rmax＞RWSL；
At OP1Marking point P in clockwise direction, marking radius OP in interface1The intersection point OP of the arc and the line segment OP2Wherein OP1＝OP2＝RmaxRequires P1、P2The circular arc segment between should contain complete NWLWhite target bar, default NWL＝5；
Radius OP1、OP2And P1P2The arc segment defines a sector OP to be detected1P2。
5. The automatic detection method according to claim 1, wherein the extraction of the data of the area to be detected specifically comprises:
calculating the number of rows and columns of the data matrix: according to sector area OP to be detected1P2Maximum radius R ofmaxMinimum radius RminAnd the radius stepping value delta r is summed, and the total row number M of the data matrix is obtained by calculationf＝1+(Rmax-Rmin) A/Δ R, wherein Rmax、RminThe unit of Δ r is pixel; according to the left edge angle of the detection area
Figure FDA0002646590530000021
I.e. radius OP1Corresponding angle, right sideEdge angle
Figure FDA0002646590530000022
I.e. radius OP2Corresponding angle and angle step value
Figure FDA0002646590530000023
Calculating the total number of columns of the target data matrix
Figure FDA0002646590530000024
Wherein the content of the first and second substances,
Figure FDA0002646590530000025
Figure FDA0002646590530000026
the unit of (a) is an angle;
data matrix row data extraction: a circular arc segment corresponding to the radius R of the target, wherein Rmin≥r≥Rmin. Firstly, according to a nearest principle, namely selecting a pixel closest to the calculated coordinate distance, determining a gray value of a leftmost end point of the arc segment, and filling the gray value into the 1 st row and the 1 st column of the data matrix; then is provided with
Figure FDA0002646590530000027
Determining the gray value of each point of the arc segment to the right according to the nearest principle as an angle stepping value, and filling the gray value into each column of the ith row of the data matrix in sequence; and
and (3) data matrix composition: from i to M in sequencefCarrying out data extraction on each data line of the data matrix in rows to finally form an MfLine by NfA data matrix of columns; each row of the data matrix corresponds to a different target radius.
6. The automatic detection method of claim 1, wherein the initial row detection is performed by traversing each row mean in the acquired data matrix, and the initial row corresponds to a first row having a row mean greater than a background black level.
7. The automatic detection method of claim 1, the extreme radius detection comprising:
white line detection; and
selecting a minimum threshold; and/or
The white line detection is carried out by adopting a maximum value identification algorithm, the gray change of each row of data in the data matrix is analyzed line by line, the number of the gray maximum values in the data is detected, and each time one gray maximum value is detected, a white target strip is considered to be detected; and/or
The threshold value used for the white line detection judgment needs to accord with the human eye perception rule, is the lowest limit of visual perception, and needs to be determined according to the lowest distinguishable contrast of human eyes under different background conditions.
8. The automatic detection method according to claim 1, wherein the calculation formula for calculating the ground resolution is as follows:
Figure FDA0002646590530000031
wherein D isWSLRepresenting the outer ring chord length; according to the Nyquist sampling theory, the minimum distinguishable line width should not be less than 1 ground sampling pixel (GSD), which requires a limiting radius
Figure FDA0002646590530000032
So when R isminWhen the resolution is 0, the calculation formula of the ground resolution of the external field test remote sensing image is shown as the following formula:
Figure FDA0002646590530000033
9. a computer device, comprising:
one or more processors;
a memory to store one or more instructions that,
wherein the one or more instructions, when executed by the one or more processors, cause the one or more processors to implement the method of any one of claims 1-8.
10. A computer readable storage medium having stored thereon executable instructions which, when executed by a processor, cause the processor to carry out the method of any one of claims 1 to 8.
CN202010860235.3A 2020-08-24 2020-08-24 Optical remote sensing load ground resolution automatic detection method based on radial target Active CN111932526B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010860235.3A CN111932526B (en) 2020-08-24 2020-08-24 Optical remote sensing load ground resolution automatic detection method based on radial target

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010860235.3A CN111932526B (en) 2020-08-24 2020-08-24 Optical remote sensing load ground resolution automatic detection method based on radial target

Publications (2)

Publication Number Publication Date
CN111932526A true CN111932526A (en) 2020-11-13
CN111932526B CN111932526B (en) 2022-12-06

Family

ID=73306107

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010860235.3A Active CN111932526B (en) 2020-08-24 2020-08-24 Optical remote sensing load ground resolution automatic detection method based on radial target

Country Status (1)

Country Link
CN (1) CN111932526B (en)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080088719A1 (en) * 2005-04-29 2008-04-17 Eliezer Jacob Digital camera with non-uniform image resolution
CN102749623A (en) * 2012-07-13 2012-10-24 中国资源卫星应用中心 Method for testing ground sampling distance of high-accuracy remote sensing satellite based on target
CN104764465A (en) * 2015-03-30 2015-07-08 中国资源卫星应用中心 Method for measuring ground resolution of remote sensing satellite

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080088719A1 (en) * 2005-04-29 2008-04-17 Eliezer Jacob Digital camera with non-uniform image resolution
CN102749623A (en) * 2012-07-13 2012-10-24 中国资源卫星应用中心 Method for testing ground sampling distance of high-accuracy remote sensing satellite based on target
CN104764465A (en) * 2015-03-30 2015-07-08 中国资源卫星应用中心 Method for measuring ground resolution of remote sensing satellite

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
彭宇等: "基于辐射状靶标的天绘一号卫星CCD相机分辨率在轨检测", 《测绘与空间地理信息》 *

Also Published As

Publication number Publication date
CN111932526B (en) 2022-12-06

Similar Documents

Publication Publication Date Title
JP6620477B2 (en) Method and program for detecting cracks in concrete
JP7319432B2 (en) LEARNING DATA COLLECTION DEVICE, LEARNING DATA COLLECTION METHOD, AND PROGRAM
JP6099479B2 (en) Crack detection method
JP4811567B2 (en) Stress measurement method for structures using photographed images
US20160133008A1 (en) Crack data collection method and crack data collection program
JP2021060656A (en) Road damage determination device, road damage determination method, and road damage determination program
CN110211185B (en) Method for identifying characteristic points of calibration pattern in group of candidate points
CN103389310B (en) Online sub-pixel optical component damage detection method based on radiation calibration
WO2021010269A1 (en) Inspection device, inspection method, and program
JP2011242365A (en) Crack detection device and crack detection method for concrete structure image, and program therefor
JP5852919B2 (en) Crack detection method
JP5705711B2 (en) Crack detection method
JP4870016B2 (en) Crack detection method
JP5957250B2 (en) Crack detection method and crack display device
CN114419045A (en) Method, device and equipment for detecting defects of photoetching mask plate and readable storage medium
CN111932526B (en) Optical remote sensing load ground resolution automatic detection method based on radial target
CN117115194B (en) Contour extraction method, device, equipment and medium based on electron microscope image
CN114187363A (en) Method and device for obtaining radial distortion parameter value and mobile terminal
Calderón et al. Crack detection in concrete elements from RGB pictures using modified line detection kernels
CN110809779A (en) Image synthesizing method, image synthesizing apparatus, and recording medium
CN113240635B (en) Structural object detection image quality testing method with crack resolution as reference
JP5642605B2 (en) Inspection apparatus, program, and image alignment method
EP3582179B1 (en) Image processing apparatus, image processing method, and image processing program
JP2004134861A (en) Resolution evaluation method, resolution evaluation program, and optical apparatus
JP6909090B2 (en) Quantitative evaluation method, quantitative evaluation device, quantitative evaluation system and program of damaged part on concrete surface

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant