+ All Categories
Home > Documents > Method of rotation angle measurement in machine vision based on calibration pattern with spot array

Method of rotation angle measurement in machine vision based on calibration pattern with spot array

Date post: 05-Oct-2016
Category:
Upload: bin
View: 215 times
Download: 0 times
Share this document with a friend
6
Method of rotation angle measurement in machine vision based on calibration pattern with spot array Weimin Li,* Jing Jin, Xiaofeng Li, and Bin Li Department of Precision Machinery and Precision Instrumentation, University of Science and Technology of China, 96#, Jing Zhai Road, Hefei, Anhui Province, China *Corresponding author: [email protected] Received 11 September 2009; revised 9 December 2009; accepted 20 January 2010; posted 25 January 2010 (Doc. ID 116185); published 17 February 2010 We propose a method of rotation angle measurement with high precision in machine vision. An area scan CCD camera, imaging lens, and calibration pattern with a spot array make up the measurement device for measuring the rotation angle. The calibration pattern with a spot array is installed at the rotation part, and the CCD camera is set at a certain distance from the rotation components. The coordinates of the spots on the calibration pattern is acquired through the vision image of the calibration pattern cap- tured by the CCD camera. At the initial position of the calibration pattern, the camera is calibrated with the spot array; the mathematical model of distortion error of the CCD camera is built. With the equation of coordinate rotation measurement, the rotation angle of the spot array is detected. In the theoretic simulation, noise of different levels is added to the coordinates of the spot array. The experiment results show that the measurement device can measure the rotation angle precisely with a noncontact method. The standard deviation of rotation angle measurement is smaller than 3 arc sec. The measurement device can measure both microangles and large angles. © 2010 Optical Society of America OCIS codes: 120.3930, 150.1488, 150.1135, 350.4600, 330.7325, 330.1880. 1. Introduction At present, rotation angle measurement is an impor- tant component of geometrical measurement tech- nology and has been developed completely [14]. Measurement accuracy has reached a high level. Now the main task is to improve accuracy and reso- lution of measurement. Researchers pay close attention to the optical an- gle-measurement method, which is noncontact and provides high accuracy and high sensitivity. In par- ticular, the development of a stable laser source makes it possible to use it in industrial field mea- surements. As a result, the optical rotation-angle- measurement method is being used more and more widely, and various new optical rotation angle mea- surement methods have emerged. At present, be- sides the optical dividing head method and the polygon method, optical rotation angle measurement methods also include the optical encoder method, the diffraction method, the autocollimation method, the fiber method [5], the acousto-optic modulation meth- od, the circular grating method, the optical internal- reflection method, the laser interference method, the parallel interferogram method, and the ring laser method. Many of these methods have been applied successfully in the precision measurement of small angles and they share the merits of high accuracy and sensitivity. Currently, the circular grating, optical internal- reflection, laser interference, ring laser [6], and vision angle measurement methods are based on straight lines [7]. Li and Yu [4] have used the vision method to mea- sure angles in-plane. They proposed an angle- measurement method based on image, which uses a straight line as a characteristic symbol. However, 0003-6935/10/061001-06$15.00/0 © 2010 Optical Society of America 20 February 2010 / Vol. 49, No. 6 / APPLIED OPTICS 1001
Transcript
Page 1: Method of rotation angle measurement in machine vision based on calibration pattern with spot array

Method of rotation angle measurement in machinevision based on calibration pattern

with spot array

Weimin Li,* Jing Jin, Xiaofeng Li, and Bin LiDepartment of Precision Machinery and Precision Instrumentation, University of Science

and Technology of China, 96#, Jing Zhai Road, Hefei, Anhui Province, China

*Corresponding author: [email protected]

Received 11 September 2009; revised 9 December 2009; accepted 20 January 2010;posted 25 January 2010 (Doc. ID 116185); published 17 February 2010

We propose a method of rotation angle measurement with high precision in machine vision. An area scanCCD camera, imaging lens, and calibration pattern with a spot array make up the measurement devicefor measuring the rotation angle. The calibration pattern with a spot array is installed at the rotationpart, and the CCD camera is set at a certain distance from the rotation components. The coordinates ofthe spots on the calibration pattern is acquired through the vision image of the calibration pattern cap-tured by the CCD camera. At the initial position of the calibration pattern, the camera is calibrated withthe spot array; the mathematical model of distortion error of the CCD camera is built. With the equationof coordinate rotation measurement, the rotation angle of the spot array is detected. In the theoreticsimulation, noise of different levels is added to the coordinates of the spot array. The experiment resultsshow that the measurement device can measure the rotation angle precisely with a noncontact method.The standard deviation of rotation angle measurement is smaller than 3arc sec. The measurementdevice can measure both microangles and large angles. © 2010 Optical Society of America

OCIS codes: 120.3930, 150.1488, 150.1135, 350.4600, 330.7325, 330.1880.

1. Introduction

At present, rotation angle measurement is an impor-tant component of geometrical measurement tech-nology and has been developed completely [1–4].Measurement accuracy has reached a high level.Now the main task is to improve accuracy and reso-lution of measurement.Researchers pay close attention to the optical an-

gle-measurement method, which is noncontact andprovides high accuracy and high sensitivity. In par-ticular, the development of a stable laser sourcemakes it possible to use it in industrial field mea-surements. As a result, the optical rotation-angle-measurement method is being used more and morewidely, and various new optical rotation angle mea-surement methods have emerged. At present, be-

sides the optical dividing head method and thepolygon method, optical rotation angle measurementmethods also include the optical encoder method, thediffraction method, the autocollimation method, thefiber method [5], the acousto-optic modulation meth-od, the circular grating method, the optical internal-reflection method, the laser interference method, theparallel interferogram method, and the ring lasermethod. Many of these methods have been appliedsuccessfully in the precision measurement of smallangles and they share the merits of high accuracyand sensitivity.

Currently, the circular grating, optical internal-reflection, laser interference, ring laser [6], andvision angle measurement methods are based onstraight lines [7].

Li and Yu [4] have used the vision method to mea-sure angles in-plane. They proposed an angle-measurement method based on image, which usesa straight line as a characteristic symbol. However,

0003-6935/10/061001-06$15.00/0© 2010 Optical Society of America

20 February 2010 / Vol. 49, No. 6 / APPLIED OPTICS 1001

Page 2: Method of rotation angle measurement in machine vision based on calibration pattern with spot array

there is a restriction in the method that the anglesbetween direction of the straight line and the sam-pling direction cannot be 0° and 45°. In addition,the measurement must further consider the prob-lems of lens distortion when measuring the anglesprecisely in a wide rotation range. The lens distortionof the vision measurement device is not corrected inthis method, so the measurement cannot reach highprecision when the rotation angles are large.To solve the problems in the measurement device,

it is necessary to eliminate the distortion error whenthe rotation angles are in a wide rotation range, sothis paper presents a high-precision angle measure-ment method of vision measurement based on a spotarray [8,9].

2. Principle of Vision Angle-Measurement Based onSpot Array

The angle-measurement device with machine visionis made up of an area scan CCD camera and a cali-bration pattern with a spot array, as shown in Fig. 1.The calibration pattern with spot array is put on theend face of the measured rotating pattern. The de-tecting CCD is fixed in a location that is a certain dis-tance from the measured rotating pattern. The grayimages of the spot array on the calibration patternareis captured by the CCD camera, and then the de-tection algorithm of the position of the points is usedto yield the image plane coordinates of the targetpoints. Next, the projection model of the CCD camerais built to modify the errors of lens distortion by ca-librating the camera. Then the object plane coordi-nates of the spot array are acquired. With thetransformation equations to yield the rotating angleof the spot array, the rotation angle in-plane of themeasured object is obtained.In the experiment, the CCD camera is 2000 × 2000

pixels, with a single pixel size of 7:4 μm× 7:4 μm anda focus of the optical lens of 50mm, which affects thework distance, because the image of the spot arrayshould be in the center of the field of view. Ourcalibration pattern is 140mm × 140mm, so the workdistance is about 900mm.

A. Structure of the Calibration Pattern

The glass pattern is processed to be a light transmis-sion circular hole array by semiconductor etching.The distance between adjacent holes is 5mm, and

the horizontal and vertical directions are orthogonal.The precision of the holes is 0:001mm. The holes areilluminated by the LED light source from the back ofthe calibration pattern. There are 29 × 29 holes onthe pattern. The spot array of the calibration patternis shown in Fig. 2.

B. Obtain the Image Coordinates of the Spot Array

The spot array of the calibration pattern will be im-aged on the image plane of the CCD camera throughthe imaging lens. The gray images of the spot arrayare acquired by CCD camera. The gray weightedmethod is used to measure the characteristic pointsthat represent the positions of the spots on the imageplane of the CCD camera.

The CCD sensor transforms the optical image sig-nals to gray image signals. For the spots that arewith-in the Cartesian coordinate system, a photoelectricsensor acquires the light intensity distributionVði; jÞ,and the computer can acquire the two-dimensionaldigital images Vði; jÞ of the spots. The gray signalsf ði; jÞ of each pixel ði; jÞ on the images correspondwiththe light intensity distribution Vði; jÞ of the spotimages on the photosensitivity plane of the CCD sen-sor, where ði; jÞ stands for the location order of thephotosensitivity pixels on the photosensitivity plane.ði; jÞ, which corresponds to the x, y coordinates of theCartesian coordinate system, stands for the numberof rows and columns of the pixel coordinate locationsof the corresponding points on the area scan CCD.

The gray weighted center of the spots is calculatedto represent the position of the spots from the digitalgray image:

Xc ¼

P

i;jf ði; jÞ · i

P

i;jf ði; jÞ ; ð1Þ

Yc ¼

P

i;jf ði; jÞ · j

P

i;jf ði; jÞ : ð2Þ

Fig. 1. Schematic diagram of in-plane angle measurement devicewith machine vision based on spot array.

Fig. 2. Schematic diagram of the calibration pattern with spotarray.

1002 APPLIED OPTICS / Vol. 49, No. 6 / 20 February 2010

Page 3: Method of rotation angle measurement in machine vision based on calibration pattern with spot array

Thus, the image plane coordinates of the 29 × 29spots is acquired by a gray weighted algorithm.

C. Adjustment of the CCD camera

Because of the distortions of the optical imaging sys-tem of the CCD camera [2], there is a certain degreeof distortion on the image coordinates of the spotarray of the calibration pattern. If the image coordi-nates of the spot array are used to calculate the an-gles directly, it is difficult to measure the rotationangle with high precision. The mathematical modelof the distortions of the CCD camera is created bycamera calibration to improve the detection accuracyof the measurement device.Before angle-measurement, the rotating parts are

fixed at a certain initial angle position, the object co-ordinate system, UOV, can be built up at the initialposition. The image coordinate and the object coordi-nate of the spot array at the initial angle position areused to build up the projection relationship betweenthe object coordinate system UOV and the image co-ordinate system XOY [3].For any point on the calibration pattern, object

plane coordinate ðu; vÞ can be expressed by imageplane coordinate ðx; yÞ as Eqs. (3) and (4), determinedbased on the least squares method [10,11]:

u ¼ a0 þ a1xþ a2yþ a3x2 þ a4y2 þ a5xyþ a6x3

þ a7y3 þ a8x2yþ a9xy2 þ a10x4 þ a11y4

þ a12xy3 þ a13x3yþ a14x2y2; ð3Þ

v ¼ b0 þ b1xþ b2yþ b3x2 þ b4y2 þ b5xyþ b6x3

þ b7y3 þ b8x2yþ b9xy2 þ b10x4 þ b11y4

þ b12xy3 þ b13x3yþ b14x2y2: ð4ÞEquations (3) and (4) are used to build the relation-ship between the object plane and the image plane bythe projection model. Thus, for the object coordinatesand image coordinates of the 841 standard points inthe field of view, 841 groups of this equation can be

listed. Then the nonlinear least squares method isused to yield the solution of the equation parameters(a0;a1;…;a14, b0; b1;…; b14). So the projection rela-tionship between the object plane and the imageplane is created.

The quartic equation fitting method is used to getthe parameters of the conversion equation from im-age coordinate system XOY to object coordinate sys-tem UOV. These parameters contain the lensgeometric distortions and the slope of the CCD sen-sor with the optical axis of the lens. After the camerais adjusted, the average position residual error of the841 points on the calibration pattern is 1.1272% CCDpixel and the standard deviation of the residual erroris 0.9079% CCD pixel.

D. Measuring the Object Coordinates of the Spot Array

After the calibration of the camera with the CCDcamera vision measurement device in the initialposition, the calibration equation parameters(a0;a1;…;a14, b0; b1;…; b14) are obtained and the ob-ject coordinates of the spot array on the calibrationpattern are calculated.

Fig. 3. Schematic of the rotation of the calibration pattern spotarray.

Table 1. Measurement Error of Angle under the Circumstance thatthe Object Coordinates of the Spot Array are Blurred with Noises ofDifferent Standard Deviation When the Rotation Angle is 1 arc sec

Standard Deviationof Noise

(CCD pixel)

Mean Valueof RotationAngle Error(arc second)

Standard Deviationof RotationAngle Error(arc second)

0 0 00.01 −0:013 0.1540.02 0.051 0.3080.03 −0:004 0.4530.04 0.020 0.6630.05 0.080 0.7310.06 −0:164 0.8300.07 −0:290 1.1220.08 0.289 1.4410.09 0.053 1.3410.10 −0:129 1.451

Table 2. Measurement Error of Angle under the Circumstance thatthe Object Coordinates of the Spot Array are Blurred with Noises of

Different Standard Deviation when the Rotation Angle is 45°

Standard Deviationof Noise

(CCD pixel)

Mean Valueof RotationAngle Error(arc second)

Standard Deviationof RotationAngle Error(arc second)

0 0 00.01 −0:005 0.1560.02 −0:027 0.2860.03 −0:015 0.4770.04 0.031 0.6710.05 0.100 0.7530.06 0.105 0.9030.07 0.010 1.0460.08 −0:102 1.1720.09 −0:079 1.3300.10 0.018 1.767

20 February 2010 / Vol. 49, No. 6 / APPLIED OPTICS 1003

Page 4: Method of rotation angle measurement in machine vision based on calibration pattern with spot array

When the rotating parts rotate from initial posi-tion I to position II, the camera gets the image ofthe spot array of the calibration pattern, as shownin Fig. 3. At position I, the image coordinates ofthe spot array obtained by the camera are ðx1i; y1iÞ(i ¼ 1; 2;…; 841). At position II, the image coordi-nates are ðx2i; y2iÞ (i ¼ 1; 2;…; 841). Then, using cali-bration Eqs. (3) and (4), the object plane coordinatesðu1i; v1iÞ, ðu2i; v2iÞ (i ¼ 1; 2;…; 841) of the standardpoints at position I and position II are calculated.

E. Calculation of Rotation Angle

The original object coordinates of the spot arrayðu1i; v1iÞ (i ¼ 1; 2;…; 841) and the object coordinateðu2i; v2iÞ (i ¼ 1; 2;…; 841) at the second positionof the angle are substituted into the rotation-angle-measurement Eqs. (5) and (6). With the nonlinearleast squares method, the rotation angle α of thecalibration pattern can be calculated, in whichi ¼ 1; 2;…; 841:

u2i − u20 ¼ ðu1i − u10Þ × cos α − ðv1i − v10Þ × sin α; ð5Þ

v2i − v20 ¼ ðu1i − u10Þ × sin αþ ðv1i − v10Þ × cos α; ð6Þin which ðu10; v10Þ is the center coordinate of calibra-tion pattern at initial position I and ðu20;v20Þ is thecenter coordinate of calibration pattern at position II.

3. Error Analysis

The rotation-angle-measurement accuracy of visionmeasurement based on a spot array is analyzed.

First, the theoretical simulation is realized. Then,the noise of the measured points is added to the ob-ject coordinates, and the measurement error of angledetection is acquired.

A. Impact of Measurement Errors of Points

In the experiment of simulation, the coordinateðu2i; v2iÞ (i ¼ 1; 2;…; 841) of the spots at the secondangle position is blurred with Gaussian noise. Themean of the noise is zero; the standard deviationof the noise increases from 0.01 CCD pixel to 0.1CCD pixel. With the increase of the standard devia-tion of the noise, the rotation-angle-measurement er-ror of the smaller rotation angle and the largerrotation angle are shown in Tables 1 and 2.

The error analysis shows that, when the measure-ment error of the object coordinates (standard devia-tion) is 0.03 CCD pixels, the angle measuring error(standard deviation) is 0:45 arc sec.

B. Impact of Full Rotation Angle Measurement

The object coordinate ðu2i; v2iÞ is blurred with Gaus-sian noise. Themean value of noise is 0.03 CCD pixel,and the standard deviation of noise is 0.03 CCDpixels. The spot array is rotated from 0° to 360°and the angle with the rotation of the coordinate iscalculated. The simulation of the measurementerror is shown in the Figs. 4 and 5.

The method of coordinate rotation can be used torealize accurate rotation angle measurement in thewhole angle range of 0°–360°.

Fig. 4. Average of measurement error of angle in the simulationexperiment, when the rotation angle is 1°–360°.

Fig. 5. Standard deviation of the measurement error of angle inthe simulation experiment, when the rotation angle is 1°–360°.

Table 3. Simulation Experiment of the Resolution of Rotation Angle Measurement

Rotation Angle

Measurement Errorof Rotation Angle

(arc second) Rotation Angle

Measurement Errorof Rotation Angle

(arc second) Rotation Angle

Measurement Errorof Rotation Angle

(arc second)

00°0000000 −0:055 22°3000000 −0:065 45°0000000 0.05800°0000:200 −0:014 22°3000:200 −0:022 45°0000:200 0.07200°0000:400 0.115 22°3000:400 0.078 45°0000:400 0.04900°0000:600 0.005 22°3000:600 0.050 45°0000:600 −0:00800°0000:800 −0:045 22°3000:800 −0:064 45°0000:800 −0:04800°0001:000 −0:048 22°3001:000 −0:016 45°0001:000 0.00100°0001:200 0.012 22°3001:200 −0:046 45°0001:200 0.08700°0001:400 −0:080 22°3001:400 0.078 45°0001:400 0.05600°0001:600 −0:042 22°3001:600 −0:040 45°0001:600 0.00100°0001:800 0.008 22°3001:800 0.004 45°0001:800 0.04200°0002:000 0.053 22°3002:000 −0:037 45°0002:000 −0:112

aRotation angle measurement error when the measurement error of the image coordinates of the spot array is 0.03 pixel.

1004 APPLIED OPTICS / Vol. 49, No. 6 / 20 February 2010

Page 5: Method of rotation angle measurement in machine vision based on calibration pattern with spot array

C. Resolution of Rotation Angle Measurement

The CCD camera in this system is 2000 × 2000 pixelsand the size of one pixel is 7:4 μm × 7:4 μm. As thefield of view is about 250mm × 250mm, the scopeof the object plane that a single pixel can indicateis 125 μm× 125 μm. The precision of the image planecoordinates measurement of the array is about 0.03pixel (3:75 μm). 3:75 μm stands for 6:1879 arc sec inthe field of view, so the resolution of the angle-measurement device is about 0:2arc sec becausethe spot array is 29 × 29.When the rotation components rotate a fraction of

an arc second, this rotation-angle-measurementmethod can realize the resolution of rotation anglemeasurement in the arc second grade. The measure-ment errors of the rotation angle of the spot array areshown in the Table 3, under the conditions of 0°,22:5°, and 45°, with the step size of angle rotationat 0:2 arc sec.

D. Impact of Radial Flop of the Rotation Part

When the in-plane rotation is measured, the rotationangle measurement is influenced by the radial flop of

the rotation. In the simulation experiment, the objectcoordinate of the spot array at the second angle posi-tion, ðu2i; v2iÞ (i ¼ 1; 2;…; 841), is blurred with theGaussian noise. The standard deviation of noise is0.03 CCD pixel and the mean value of noise is in-creased from 0.01 CCD pixels to 0.1 CCD pixels.The mean value of noise is equivalent to the shiftof the calibration pattern along the radial direction.Tables 4 and 5 show the angle measurement errors ofthe smaller angle and the larger angle when the ob-ject coordinate is blurred with noise of different meanvalues.

The result of simulation shows that the rotation-angle-measurementmethodwith the coordinate rota-tion of a spot array is not sensitive to such radial flop.The radial flop of the calibrationpatterndoesnothavethe obvious impact on themean value and calibrationdeviation of rotation-angle-measurement error.

4. Experimental Results

A real rotation-angle-measurement experiment isdesigned to check the effect of vision rotation anglemeasurement with a spot array. The calibrationpattern with the spot array is fixed on the rotary ta-ble, whose angle rotation precision is 2 arc sec. Thereal rotation angle of calibration pattern is θ. The ro-tation angle measured by the vision measurementdevice is θ1. The error of rotation angle measurementis θ − θ1. The experimental results of a small rotationangle and a large rotation angle are shown in Tables 6

Table 4. Rotation Measurement Errors When there is Noise withDifferent Mean Value in the Theoretical Simulation, under the

Condition of Tiny Rotation Angle (1 arc sec)

Mean Valueof Noise

(CCD pixel)

Mean Valueof RotationAngle Error(arc second)

Standard Deviationof RotationAngle Error(arc second)

0 0.043 0.4080.01 −0:048 0.4910.02 0.034 0.4800.03 −0:075 0.4830.04 0.004 0.4930.05 0.112 0.4600.06 0.009 0.4410.07 −0:059 0.4990.08 −0:042 0.4550.09 0.004 0.4560.10 −0:075 0.420

Table 5. Rotation Measurement Errors When there is Noise withDifferent Mean Value in the Theoretical Simulation, under the

Condition of Large Rotation Angle (45°)

Mean Valueof Noise

(CCD pixel)

Mean Valueof RotationAngle Error(arc second)

Standard Deviationof RotationAngle Error(arc second)

0 −0:018 0.3830.01 −0:089 0.4960.02 0.0389 0.4830.03 −0:0069 0.4660.04 −0:0449 0.4980.05 0.0809 0.4450.06 0.066 0.4360.07 0.007 0.4520.08 −0:021 0.4780.09 0.031 0.4800.10 0.040 0.477

Table 6. Detection Results of Tiny Rotation Angle

RotationAngle

MeasurementError of

Rotation Angle(arc second)

RotationAngle

MeasurementError of

Rotation Angle(arc second)

00°0001000 0.09 00°0201000 1.2000°0001900 −1:13 00°0201900 −1:0200°0002800 −2:96 00°0203000 −0:7800°0004200 0.60 00°0204200 −0:4300°0005400 0.68 00°0205200 −0:7600°0100300 −1:20 00°0300100 −2:7300°0101500 −0:67 00°0301100 −3:1100°0102700 0.59 00°0302600 0.8800°0103600 −1:47 00°0303500 −0:3900°0104500 −2:90 00°0304500 −1:3900°0105800 −0:56 00°0305500 −1:42

Table 7. Detection Results of Large Rotation Angle

Rotation AngleMeasurement Error

of Rotation Angle (arc second)

5°2305700 2.6210°4704500 1.0915°1704400 1.8920°4104100 1.2925°1102600 −2:0930°3503300 1.4735°0502100 −1:8640°2901200 −0:0445°5300100 −1:47

20 February 2010 / Vol. 49, No. 6 / APPLIED OPTICS 1005

Page 6: Method of rotation angle measurement in machine vision based on calibration pattern with spot array

and 7. The experimental results show that the stan-dard deviation of rotation angle measurement is lessthan 2 arc sec under small angle rotation and largeangle rotation, and that this method is an effectiveand high-precision algorithm tomeasure the rotationthat can meet the requirements of engineering appli-cations. In comparison, the standard deviation of ro-tation angle measurement is less than 5 arc sec inthe vision angle-measurement method based on astraight line, as in Ref. [4].

5. Conclusion

Compared with existing measurement techniques,this device with vision measurement can realizenoncontact rotation angle measurement. The mea-surement error of the rotating angle is less than3 arc sec (σ), and the measurement range is between0° and 360°. The self-calibration of the measurementsystem is realized and the error impact of projectiongeometric distortion is corrected. The CCD sensorand the optical axis of the lens does not need to beperpendicular. The environmental requirementsare not stringent. The method has broad applicationprospects, and has potential applications in industryand military. This device is not sensitive to the mea-surement of the spot array, flop of the rotation center,or to long periods of temperature change.

This study is supported by project “Research of keytechnology of the coordinate measurement of multi-object in large scale field,” National Natural ScienceFoundation of China (NSFC) (code 50705091).

References

1. X. Dai, O. Sasaki, J. E. Greivenkamp, and T. Suzuki, “Highaccuracy wide range rotation angle measurement by theuse of two parallel interference patterns,” Appl. Opt. 36,6190–6195 (1997).

2. W. Wang, G. Liu, Z. Pu, and S. Chen, “Measurement oftwo-dimensional small angle by means of matrix CCD,” Semi-conductor Optoelectron. 25, 134–138 (2004).

3. C. Kuang, Q. Feng, B. Zhang, Z. Zhang, and S. Chen,“Measurement method of the roll angle,” Proc. SPIE 6150,61502F (2006).

4. L. Li, Q. Yu, Z. Lei, and J. Li, “High-accuracy measurement ofrotation angle based on image,” Acta Opt Sin. 25, 491–496(2005).

5. Y. N. Kuichin, O. B. Vitrik, and A. V. Dyshlyuk, “Stabilizedfiber optic sensor for remote measuring angle of inclination,”Proc. SPIE 6595, 65952G (2007).

6. C.-H. Liu, W.-Y. Jywe, and S. C. Tzeng, “Simple three-dimensional laser angle sensor for three-dimensional small-angle measurement,” Appl. Opt. 43, 2840–2845 (2004).

7. J. A. Muñoz-Rodríguez, A. Asundi, and R. Rodriguez-Vera,“Recognition of a light line pattern by Hu moments for 3-Dreconstruction of a rotated object,” Opt. Laser Technol. 37,131–138 (2005).

8. Z. Li, Y. Shi, C. Wang, and Y. Wang, “Accurate calibrationmethod for a structured light system,” Opt. Eng. 47, 053604(2008).

9. X. Chen, J. Xi, Y. Jin, and J. Sun, “Accurate calibration for acamera–projector measurement system based on structuredlight projection,” Opt. Lasers Eng. 47, 310–319 (2009).

10. J. A. Muñoz-Rodríguez, “Shape connection by pattern recogni-tion and laser metrology,” Appl. Opt. 47, 3590–3608 (2008).

11. J. A. Muñoz-Rodríguez, “Modeling of a mobile setup bynetworks for object contouring,” Opt. Eng. 47, 053605 (2008).

1006 APPLIED OPTICS / Vol. 49, No. 6 / 20 February 2010


Recommended