CN107192348A - A kind of high-precision 3D vision measuring methods - Google Patents
A kind of high-precision 3D vision measuring methods Download PDFInfo
- Publication number
- CN107192348A CN107192348A CN201610140647.3A CN201610140647A CN107192348A CN 107192348 A CN107192348 A CN 107192348A CN 201610140647 A CN201610140647 A CN 201610140647A CN 107192348 A CN107192348 A CN 107192348A
- Authority
- CN
- China
- Prior art keywords
- camera
- precision
- parameter
- edge
- pixel
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B11/00—Measuring arrangements characterised by the use of optical techniques
- G01B11/24—Measuring arrangements characterised by the use of optical techniques for measuring contours or curvatures
Abstract
The invention provides a kind of high-precision 3D vision measuring methods.This method carries out three-dimensional measurement to object with rebuilding using three pairs of cubic phase machine equipments, a series of problems, such as traditional 2D vision measuring methods can not measure object height, depth, thickness, flatness is this method solve, and three-dimensional measurement precision has reached 10um.
Description
Technical field
The present invention relates to a kind of high-precision 3D vision measuring methods, belong to machine vision, three-dimensional reconstruction and high-accuracy automatic
Change measuring instrument field.
Background technology
In manufacturing industry, certain accuracy of manufacture error is inevitably there is in itself for the machine that produces workpiece.
Therefore, every workpiece produced by it will necessarily have certain tolerance.For those requirement on machining accuracy very high workpiece
For, we must accurately measure the poor scope of its work.
Vision measurement has also embodied its unique advantage in terms of the dimensional measurement of industrial part, and noncontact inspection can be achieved
Measurement, and long-time stable work.Using vision measurement, a large amount of human resources can be saved, it is considerable to be that company brings
Benefit, its application prospect is considerable.Vision measurement system has measurement function, is capable of the apparent size of automatic measurement product,
Than if the size of the parameter such as appearance profile, aperture, area of automatic measurement product.
The current detection means in field of machine vision main flow relies primarily on the vision measurement technology of 2D cameras.This side
Method can only measure the correspondingly-sized for the object for belonging to 2D, when needs measure height, depth, thickness, flatness, volume, abrasion
When spending, the measuring method is then unable to reach requirement.Industrially, the instrument for high-precision part dimensional measurement is three-dimensional coordinate
Measuring instrument, its precision is general all in micron level, but three-coordinates measuring machine measurement efficiency is low, is manufactured far from industry is met
Actual efficiency demand.In recent years, as " proposition of industry 4.0 ", the vision measurement technology being closely related with intelligent plant is obtained
Great attention is arrived and has developed.
Therefore, the high-precision 3D vision measurements technology of research and development rapidly and efficiently, it has also become realize " the eager need of industry 4.0 "
Ask.
The content of the invention
The present invention is to overcome the above-mentioned deficiency of prior art, it is proposed that a kind of high-precision 3D vision measuring methods, should
Method overcomes the problem of traditional 2D vision measurements can not measure the relevant parameter not between approximately the same plane target, while
The problem of traditional three-dimensional coordinate measuring instrument measurement efficiency is low is overcome, directly measurement target is entered by multipair binocular camera
Row 3D is measured, and precision has reached below 10um.
The present invention is realized by such technical scheme：A kind of high-precision 3D vision measuring methods, it is characterised in that pass through
The three pairs of binocular cameras in left side, right side, top for being placed in target to be measured rebuild the three-dimensional coordinate of target to be measured respectively.Realize described
Method comprises the following steps：
Step 1, high-precision calibration of camera internal parameters, determines the intrinsic parameter of all cameras.First with Zhang Zhengyou standardizations
The intrinsic parameter of camera is calculated, the pixel that variance is more than defined threshold is deleted further according to covariance law.Afterwards, remaining side is used
Difference carries out camera calibration again less than the pixel of defined threshold.
Step 2, stereo camera calibration, determines the outer parameter of stereoscopic camera.Determining the outer parameter of camera needs to use step 1
The result of the intrinsic parameter of middle demarcation, in order to eliminate the propagation effect of error, needs to carry out all result of calculation in this step
Binding collection adjusting and optimizing is calculated.
Step 3, the conversion of different cameral coordinate system, same camera coordinates system is uniformly transformed into by different cameral coordinate system
Under, the conversion of six parameter coordinate systems is carried out using Quaternion method.
Step 4, extract the marginal position of the object in each camera fields of view (by taking ellipse target as an example).Use first
Canny rim detections extract the marginal position of entire image designated area, then calculate with border following algorithm each edge picture
The number of vegetarian refreshments, deletes the edge that pixel is less than fixed threshold.
Step 5, elliptical edge in step 4 is fitted, digital simulation oval major and minor axis and area, deletion are unsatisfactory for
The elliptical edge of condition, and extract elliptical center.
Step 6, obtained respectively according to the pixel coordinate of two elliptical centers in left and right in step 5 with step 1 and step 2
Camera parameter rebuilds left side, right side, the three-dimensional coordinate of top elliptical center respectively with Linear Triangular method first, according to line
Property trigonometric ratio result nonlinear optimization is carried out to the three-dimensional point of reconstruction, obtain the higher three-dimensional coordinate point of precision.
Step 7, all three-dimensional points are transformed under same coordinate system according to obtaining coordinate system conversion parameter in step 3,
It can obtain high accuracy three-dimensional coordinate points.
The present invention has the advantage that compared with prior art：
1. the present invention directly can carry out three-dimensional measurement to target to be measured, the three-dimensional coordinate of target to be measured is obtained.
2. camera calibration precision is high, the stated accuracy of existing conventional Zhang Zhengyou standardizations is about 0.4 pixel, the present invention
Camera calibration precision be about 0.1 pixel.
3. the elliptical edge extraction accuracy of the present invention is controlled within 0.2 pixel.
4. the final reconstruction accuracy of the present invention is within 0.3 pixel.
Brief description of the drawings
Fig. 1 is the inventive method camera installation position schematic diagram.
Fig. 2 is the pictorial diagram of the inventive method object to be measured.
Fig. 3 is the inventive method algorithm flow chart.
Fig. 4 is the inventive method and the comparison diagram of Zhang Zhengyou calibration algorithm re-projection errors.
Fig. 5 is the gridiron pattern locus figure reconstructed.
Fig. 6 is that object edge extracts result.
Fig. 7 is edge fitting and its topography.
Embodiment
With reference to the accompanying drawings and examples, the present invention will be further described.
As shown in figure 1, disposing scheme for measurement process camera, wherein two cameras of left side placement, right side disposes two phases
Machine, top is installed by two cameras.Fig. 2 is the pictorial diagram of the measurement object of the present invention.The flow chart of the present invention is as shown in Figure 3.
In the present embodiment, measurement target, the measurement left side center of circle corresponding with right side are used as with the calibrating block of a manual manufacture
The distance of connected straight line, the calibrating block obtains the three-dimensional coordinate in the center of circle by three-dimensional coordinate measuring instrument, and precision is 1um.In camera
In calibration process, the glass planar gridiron pattern scaling board that the demarcation thing that we use is 10um for precision.
Below in conjunction with the accompanying drawings, the detailed process to the technical scheme of invention is illustrated.Because present invention left side, right side, on
The camera calculating process of side is identical, therefore is described by taking the calculating process of two cameras in left side as an example.
(1) high-precision intrinsic parameter is carried out to each camera of left side to demarcate, determine the Intrinsic Matrix (4 of left camera
Parameter is respectively fx, fy, cx, cy).The step calculates the Intrinsic Matrix of camera first with Zhang Zhengyou standardizations, and we obtain
5 intrinsic parameters to two cameras are respectively [10912.9 10908.1 2069.9 1018.5], [10765.4 10767.6
2030.3 1066.4], the pixel that variance is more than defined threshold is deleted further according to covariance law.Afterwards, remaining variance is used
Pixel less than defined threshold carries out camera calibration again, and the intrinsic parameter of two obtained camera is respectively [10916.4
10915.4 2030.0 1100.8]、[10893.7 10889.9 2083.5 1104.1].Fig. 4 is obtained by above-mentioned steps
Intrinsic Matrix reconstruct come three-dimensional point re-projection error distribution, it can be seen that error be substantially distributed in 0.1 pixel with
It is interior.
(2) stereo camera calibration, it is determined that the outer parameter of two, left side camera.Determine that the outer parameter of camera needs to use (1) acceptance of the bid
The result of fixed intrinsic parameter, in order to eliminate the propagation effect of error, needs to carry out binding collection to all result of calculation in this step
(or bundle adjustment) optimization calculation procedure is adjusted, obtaining outer parameter ist
=[- 168.039, -1.172,41.871].Fig. 5 is to reconstruct tessellated location drawing picture in three dimensions according to stereo calibration.
(3) extract the marginal position of the object in each camera fields of view (by taking ellipse target as an example).Canny is used first
Rim detection extracts the marginal position of entire image designated area, then calculates with border following algorithm each edge pixel point
Number, deletes the edge that pixel is less than fixed threshold.In the present embodiment, we select to delete number of edge points less than 500
Edge.Marginal position extraction effect is as shown in Figure 6.
(5) elliptical edge in (4) is fitted, digital simulation oval major and minor axis and area, deletion are unsatisfactory for condition
Elliptical edge, and extract elliptical center.In the present embodiment, five with direct method solution elliptic equation that we use
Parameter, the major and minor axis and corresponding area of ellipse can be tried to achieve according to obtained elliptic parameter, in this, as whether to retain this ellipse
The foundation of rounded edge.Final elliptical edge extracts result and its partial enlarged drawing is as shown in Figure 7.
(6) respectively according to the pixel coordinate of two elliptical centers in left and right in (5), the camera parameter obtained with (1) (2) is first
The three-dimensional coordinate of elliptical center is rebuild with Linear Triangular method, now because the presence of systematic error and random error, is obtained
The coordinate precision of the elliptical center arrived is not high.We carry out non-linear further according to Linear Triangular result to the three-dimensional point of reconstruction
Optimization, the higher three-dimensional coordinate point of precision is obtained with this.
(7) above-mentioned steps are being combined, carrying out (1)-(6) step to right side and top camera respectively calculates, and according to quaternary
All three-dimensional points are transformed under same coordinate system by plain method, you can obtain the distance of corresponding three-dimensional coordinate to be measured.With three-dimensional
The three-dimensional point coordinate that coordinate measuring apparatus is obtained is standard value, the three-dimensional point that the left perspective camera measurement that the final present invention is obtained goes out
Resultant error is analyzed as follows table 1 to coordinate in contrast, and the three-dimensional point coordinate that perspective right camera measurement goes out in contrast miss by result
Difference is analyzed as follows table 2, and resultant error is analyzed as follows table 3 to the three-dimensional point coordinate that top perspective camera measurement goes out in contrast.
Table 1
| Target | Target 1 | Target 2 | Target 3 | Target 4 | Target 5 |
| Error (um) | -1.7 | 3.6 | 1.9 | 9.3 | -5.4 |
Table 2
| Target | Target 1 | Target 2 | Target 3 | Target 4 | Target 5 |
| Error (um) | 2.7 | 3.5 | 5.5 | 6.1 | 2.2 |
Table 3
| Target | Target 1 | Target 2 | Target 3 | Target 4 | Target 5 |
| Error (um) | 3.1 | -3.5 | -1.3 | 8.5 | -5.2 |
Particular embodiments described above, has been carried out furtherly to the purpose of the present invention, technical scheme and beneficial effect
Bright, understanding used the foregoing is only the specific embodiment of the present invention, be not intended to limit the invention, all at this
Within the spirit and principle of invention, any modification, equivalent substitution and improvements done etc. should be included in the protection model of the present invention
Within enclosing.
Claims (7)
1. a kind of high-precision 3D vision measuring methods, it is characterised in that comprise the following steps：
Step 1, high-precision calibration of camera internal parameters, determines the intrinsic parameter of all cameras.Calculated first with Zhang Zhengyou standardizations
The intrinsic parameter of camera, the pixel that variance is more than defined threshold is deleted further according to covariance law.Afterwards, it is small with remaining variance
Camera calibration is carried out again in the pixel of defined threshold.
Step 2, stereo camera calibration, determines the outer parameter of stereoscopic camera.Determine that the outer parameter of camera needs to use step 1 to get the bid
The result of fixed intrinsic parameter, in order to eliminate the propagation effect of error, needs to bundle all result of calculation in this step
Collect adjusting and optimizing to calculate.
Step 3, the conversion of different cameral coordinate system, different cameral coordinate system is uniformly transformed under same camera coordinates system,
The conversion of six parameter coordinate systems is carried out using Quaternion method.
Step 4, extract the marginal position of the object in each camera fields of view (by taking ellipse target as an example).Canny is used first
Rim detection extracts the marginal position of entire image designated area, then calculates with border following algorithm each edge pixel point
Number, deletes the edge that pixel is less than fixed threshold.
Step 5, elliptical edge in step 4 is fitted, digital simulation oval major and minor axis and area, deletion are unsatisfactory for condition
Elliptical edge, and extract elliptical center.
Step 6, respectively according to the pixel coordinate of two elliptical centers in left and right in step 5, the camera obtained with step 1 and step 2
Parameter rebuilds left side, right side, the three-dimensional coordinate of top elliptical center respectively with Linear Triangular method first, according to linear three
Angling result carries out nonlinear optimization to the three-dimensional point of reconstruction, obtains the higher three-dimensional coordinate point of precision.
Step 7, all three-dimensional points are transformed under same coordinate system according to obtaining coordinate system conversion parameter in step 3, you can
Obtain high accuracy three-dimensional coordinate points.
2. a kind of high-precision 3D vision measuring methods according to claim 1, it is characterised in that the specific step of the step 1
Suddenly it is：High-precision intrinsic parameter demarcation is carried out to each camera, 4 parameters for determining the Intrinsic Matrix of left camera are respectively
fx、fy、cx、cy.The step calculates the Intrinsic Matrix of camera first with Zhang Zhengyou standardizations, and we obtain two cameras
5 intrinsic parameters are respectively to delete the pixel that variance is more than defined threshold further according to covariance law.Afterwards, remaining side is used
Difference carries out camera calibration again less than the pixel of defined threshold, the intrinsic parameter of two obtained camera.
3. a kind of high-precision 3D vision measuring methods according to claim 1, it is characterised in that the specific step of the step 2
Suddenly it is：Determine the outer parameter of each two stereoscopic camera pair.Determining the outer parameter of camera needs to use the intrinsic parameter demarcated in step 1
Result, in order to eliminate the propagation effect of error, need to carry out all result of calculation binding collection adjusting and optimizing in this step
Calculation procedure, obtains accurate outer parametric results.
4. a kind of high-precision 3D vision measuring methods according to claim 1, it is characterised in that the specific step of the step 3
Suddenly it is：Extract the marginal position of the object in each camera fields of view (by taking ellipse target as an example).Examined first with Canny edges
Survey the marginal position for extracting entire image designated area, then calculate with border following algorithm the number of each edge pixel point,
Delete pixel and be less than the edge of fixed threshold, and delete the edge that number of edge points is less than 500.
5. a kind of high-precision 3D vision measuring methods according to claim 1, it is characterised in that the specific step of the step 4
Suddenly it is：Elliptical edge in step 4 is fitted, digital simulation oval major and minor axis and area, deletion are unsatisfactory for the ellipse of condition
Rounded edge, and elliptical center is extracted, five parameters that elliptic equation is solved with direct method of use are joined according to obtained ellipse
Number can try to achieve the major and minor axis and corresponding area of ellipse, in this, as whether retaining the foundation of the elliptical edge.
6. a kind of high-precision 3D vision measuring methods according to claim 1, it is characterised in that the specific step of the step 5
Suddenly it is：Respectively according to the pixel coordinate of two elliptical centers in left and right in step 4, with the camera parameter obtained in step 1 and step 2
The three-dimensional coordinate of elliptical center is rebuild with Linear Triangular method first, now because systematic error and random error are deposited
The coordinate precision of obtained elliptical center is not high.We are carried out further according to Linear Triangular result to the three-dimensional point of reconstruction
Nonlinear optimization, the higher three-dimensional coordinate point of precision is obtained with this.
7. a kind of high-precision 3D vision measuring methods according to claim 1, it is characterised in that the specific step of the step 6
Suddenly it is：All three-dimensional points are transformed under same coordinate system according to Quaternion method, you can obtain corresponding three-dimensional coordinate to be measured
And its exact position of correspondence target.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610140647.3A CN107192348A (en) | 2016-03-14 | 2016-03-14 | A kind of high-precision 3D vision measuring methods |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610140647.3A CN107192348A (en) | 2016-03-14 | 2016-03-14 | A kind of high-precision 3D vision measuring methods |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107192348A true CN107192348A (en) | 2017-09-22 |
Family
ID=59870763
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610140647.3A Pending CN107192348A (en) | 2016-03-14 | 2016-03-14 | A kind of high-precision 3D vision measuring methods |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107192348A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108154536A (en) * | 2017-12-13 | 2018-06-12 | 南京航空航天大学 | The camera calibration method of two dimensional surface iteration |
| CN109341591A (en) * | 2018-11-12 | 2019-02-15 | 杭州思看科技有限公司 | A kind of edge detection method and system based on handheld three-dimensional scanner |
| CN111815704A (en) * | 2020-07-16 | 2020-10-23 | 中国农业机械化科学研究院 | Grain bin grain heap volume measuring and calculating method based on binocular camera |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101231750A (en) * | 2008-02-21 | 2008-07-30 | 南京航空航天大学 | Calibrating method of binocular three-dimensional measuring system |
| CN101995231A (en) * | 2010-09-20 | 2011-03-30 | 深圳大学 | Three-dimensional detection system for surface of large thin-shell object and detection method thereof |
| CN102944191A (en) * | 2012-11-28 | 2013-02-27 | 北京航空航天大学 | Method and device for three-dimensional vision measurement data registration based on planar circle target |
| CN104315978A (en) * | 2014-11-10 | 2015-01-28 | 北京理工大学 | Method and device for measuring pipeline end face central points |
| CN104809738A (en) * | 2015-05-18 | 2015-07-29 | 长春工业大学 | Airbag overall dimension detection method based on binocular vision |
| CN105043250A (en) * | 2015-05-29 | 2015-11-11 | 南京邮电大学 | Dual-view-angle data alignment method based on at least two common mark points |
| CN105157609A (en) * | 2015-09-01 | 2015-12-16 | 大连理工大学 | Two-sets-of-camera-based global morphology measurement method of large parts |
-
2016
- 2016-03-14 CN CN201610140647.3A patent/CN107192348A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101231750A (en) * | 2008-02-21 | 2008-07-30 | 南京航空航天大学 | Calibrating method of binocular three-dimensional measuring system |
| CN101995231A (en) * | 2010-09-20 | 2011-03-30 | 深圳大学 | Three-dimensional detection system for surface of large thin-shell object and detection method thereof |
| CN102944191A (en) * | 2012-11-28 | 2013-02-27 | 北京航空航天大学 | Method and device for three-dimensional vision measurement data registration based on planar circle target |
| CN104315978A (en) * | 2014-11-10 | 2015-01-28 | 北京理工大学 | Method and device for measuring pipeline end face central points |
| CN104809738A (en) * | 2015-05-18 | 2015-07-29 | 长春工业大学 | Airbag overall dimension detection method based on binocular vision |
| CN105043250A (en) * | 2015-05-29 | 2015-11-11 | 南京邮电大学 | Dual-view-angle data alignment method based on at least two common mark points |
| CN105157609A (en) * | 2015-09-01 | 2015-12-16 | 大连理工大学 | Two-sets-of-camera-based global morphology measurement method of large parts |
Non-Patent Citations (3)
| Title |
|---|
| 万刚 等: "《无人机测绘技术及应用》", 31 December 2015, 测绘出版社 * |
| 吴伟仁 等: "《深空探测器自主导航原理与技术》", 31 May 2011, 中国宇航出版社 * |
| 吴侃 等: "《矿山开采沉陷监测及预测新技术》", 31 December 2012, 中国环境科学出版社 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108154536A (en) * | 2017-12-13 | 2018-06-12 | 南京航空航天大学 | The camera calibration method of two dimensional surface iteration |
| CN109341591A (en) * | 2018-11-12 | 2019-02-15 | 杭州思看科技有限公司 | A kind of edge detection method and system based on handheld three-dimensional scanner |
| CN111815704A (en) * | 2020-07-16 | 2020-10-23 | 中国农业机械化科学研究院 | Grain bin grain heap volume measuring and calculating method based on binocular camera |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Chen et al. | High-accuracy multi-camera reconstruction enhanced by adaptive point cloud correction algorithm | |
| Garcia et al. | A combined temporal tracking and stereo-correlation technique for accurate measurement of 3D displacements: application to sheet metal forming | |
| CN105300316B (en) | Optical losses rapid extracting method based on grey scale centre of gravity method | |
| CN104359403B (en) | Planar part dimension measurement method based on sub-pixel edge algorithm | |
| CN110443836A (en) | A kind of point cloud data autoegistration method and device based on plane characteristic | |
| CN102800096B (en) | Robustness estimation algorithm of camera parameter | |
| CN109360246B (en) | Stereoscopic vision three-dimensional displacement measurement method based on synchronous subarea search | |
| CN110335234B (en) | Three-dimensional change detection method based on antique LiDAR point cloud | |
| CN105783786B (en) | Part chamfering measuring method and device based on structured light vision | |
| CN103535960B (en) | Human body three-dimensional measurement method based on digital images | |
| CN108389250A (en) | The method for quickly generating building cross-section diagram based on point cloud data | |
| CN112082493B (en) | Binocular imaging-based pipeline flange inner radius visual measurement method | |
| CN104392476A (en) | Method of extracting three-dimensional axis of tunnel based on minimum bounding box algorithm | |
| CN104748683A (en) | Device and method for online and automatic measuring numerical control machine tool workpieces | |
| CN103278138A (en) | Method for measuring three-dimensional position and posture of thin component with complex structure | |
| CN108765495B (en) | Rapid calibration method and system based on binocular vision detection technology | |
| CN104655011A (en) | Non-contact optical measurement method for volume of irregular convex-surface object | |
| CN104751458B (en) | A kind of demarcation angular-point detection method based on 180 ° of rotation operators | |
| CN101720992A (en) | Three-dimensional human body measurement method by using single camera | |
| CN107192348A (en) | A kind of high-precision 3D vision measuring methods | |
| CN109272542A (en) | A kind of determination method of three-dimension object volume | |
| CN106033653A (en) | Ground-laser-scanning-based three-dimensional indoor map making method | |
| CN106767433B (en) | A kind of method and system measuring foot sizing | |
| CN115909025A (en) | Terrain vision autonomous detection and identification method for small celestial body surface sampling point | |
| CN103679713A (en) | Two-dimensional image registration method for partially matched images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| DD01 | Delivery of document by public notice | ||
| DD01 | Delivery of document by public notice |
Addressee: Zhang Liang Document name: Notification of Passing Examination on Formalities |
|
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170922 |