Description

stereocamera to measure distance

All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.

Related Documents

Share

Transcript

Published in IET Science, Measurement and TechnologyReceived on 26th April 2008Revised on 31st August 2008doi:10.1049/iet-smt:20080055
ISSN 1751-8822
Three-dimensional measurement of distantobjects based on laser-projected CCD images
C.-C. Hsu
1
M.-C. Lu
2
W.-Y. Wang
3
Y.-Y. Lu
2
1
Department of Electrical Engineering, Tamkang University, 151 Ying-Chuan Road, Tamsui, Taipei County 25137, Taiwan,Republic of China
2
Department of Electronic Engineering, St. John’s University, 499 Tam-King Road, Section 4, Tamsui, Taipei County 25135,Taiwan, Republic of China
3
Department of Applied Electronic Technology, National Taiwan Normal University, 162 He-Ping East Road, Section 1,Taipei 10610, Taiwan, Republic of ChinaE-mail: jameshsu@mail.tku.edu.tw
Abstract:
A novel measuring system based on a single CCD camera and two laser projectors to record images andperform three-dimensional measurement of a distant object is proposed here. Because of the alignment of thelaser beams which form in parallel with the optical axis of the CCD camera, projected spots will appear on thesame scan line in a CCD image. As a result processing of a single scan line rather than the whole image is onlyrequired to identify the projected spots in the CCD image. Complex computation of video signals of the wholeimage via either pattern recognition or image analysis methods is therefore circumvented. On the basic of anestablished relationship between the distance and pixel counts between the projected spots in the CCD image, theproposed system not only measures the distance from a distant object but also the length of two arbitrarilydesignated points on the object.To provide better accuracy, intrinsic parameters of the CCD camera are taken intoconsideration in the measurement. Furthermore, the effect of laser diffusion is also proved to be irrelevant to themeasuring accuracy here. Experimental results have demonstrated that the proposed measuring method is capableof yielding accurate results of three-dimensional measurement for a distant object in a very responsive way.
1 Introduction
There are generally two ways for measuring distance: contact and non-contact approaches. The most critical defect incontact measurement is that the objects under measurement can be corrosive that inevitably causes safety concern in themeasurement. For non-contact measurement, there are various solutions to be used, like laser reﬂection [1–5] andultrasonic (or light) reﬂection [6–9] methods. Althoughtechniques based on laser and ultrasonic reﬂection have theadvantage of speed, surface reﬂectivity of the object playsan important role. If the reﬂection surface is undesired, themeasuring system generally performs poorly or not at all. We can say that poor surface reﬂectivity is the commonfailure of non-contact measurement systems. Furthermore,these methods also have difﬁculties in recording images while measuring distance of the objects. Although patternrecognition or image analysis methods [10–15] provide analternative for measuring distance, huge amount of storagecapacity and high-speed microprocessors are generally required because of the need to store the whole image for further processing. As a result, the performance of real-time measurement is generally not satisfactory because of the speed constraint. Based on triangulation relationship[16–18], an image-based distance measuring system(IBDMS) [16–21] was successfully applied to measureheight for liquid or particles in tanks. The triangular measuring method, however, suffered from a seriousdrawback of nonlinear distortion when measuring objects ina short photographing distance, which inevitably resulted ina larger measurement error. To overcome the distortionproblems [22, 23] caused by mechanisms, lens, and circuitsassociated with the CCD in the triangular measuring method, a parallel measuring method [20, 21] wasproposed, providing a simpler structure for achieving moreaccurate and real-time measurement of distance. Although
IET Sci. Meas. Technol.
, 2009, Vol. 3, Iss. 3, pp. 197–207 197doi:10.1049/iet-smt:20080055
&
The Institution of Engineering and Technology 2009
www.ietdl.org
Authorized licensed use limited to: National Central University. Downloaded on November 25, 2009 at 08:47 from IEEE Xplore. Restrictions apply.
this worked satisfactorily to some extent there is still room for further improvement on the performance of the parallelmeasuring method. For example, the distance between theoptical origin (op) and front end of the camera isunknown, resulting in a critical measurement error. Theexisting IBDMS methods also failed to consider thediffusion effect of the laser-projected spots in the images as well as the inﬂuence of lighting during the measurement.Furthermore, circuitry complexities used in existing IBDMS can be a critical drawback as far as manufacturing is concerned. Simpliﬁcation of the circuitry is also desiredfrom the viewpoint of practical applications. The rest of the paper is organised as follows. Section 2introduces an improved measuring system with superior measuring performance over existing measuring systems.Identiﬁcation of the projected spots in a video signal is givenin Section 3. Three-dimensional measurement of a distant object based on the proposed approach is given inSection 4. Experimental results of practical measurement aredemonstratedinSection5.ConclusionsaredrawninSection6.
2 The Improved parallelmeasuring system
Because of the problems mentioned earlier in existing IBDMS methods there is a need to improve theperformance of the measuring system from two aspects. The ﬁrst one is to overcome the difﬁculties in measuring a distant object situated in the measuring deadzone. Thesecond one is the determination of intrinsic parameters for various kinds of CCD cameras so as to achieve a moreaccurate measurement.
2.1 Conﬁguration of the proposed measuring system
Fig. 1 shows the conﬁguration of the proposed measuring system, where two laser projectors, Laser A and Laser B,are positioned on both sides of a CCD camera to producetwo parallel laser beams perfectly aligned with the opticalaxis of the CCD camera. Because of the disposition of Laser A, Laser B and op of the CCD shown in Fig. 2, which are positioned on a straight line, no matter how themeasuring system is rotated, the video signal of theprojected spots will appear on the same scan line. Note that the measuring system is established in such a way that theCCD camera has a distance
h
x
moved backward from themeasuring startup to overcome the difﬁculties in measuring objects situated in the deadzone, inevitably encountered via the existing approaches [21]. The distance between the opand the front end of the CCD camera
h
s
is also taken intoaccount to provide better measuring accuracy. How
h
s
isdetermined will be presented in detail later in this section. The length
D
min
between two laser projectors is ﬁxed andknown a priori. When two spots just appear on the two
Figure 1
Setup of the proposed measuring method
198
IET Sci. Meas. Technol.
, 2009, Vol. 3, Iss. 3, pp. 197–207
&
The Institution of Engineering and Technology 2009 doi:10.1049/iet-smt:20080055
www.ietdl.org
Authorized licensed use limited to: National Central University. Downloaded on November 25, 2009 at 08:47 from IEEE Xplore. Restrictions apply.
edges of the CCD image, the distance between the spots (theprojected points) and the CCD camera equals
h
x
. Themaximal pixel count in terms of external clock pulses is
N
max
.
H
k
is the distance to be measured, and
h
s
is thedistance between the op and the front end of the CCDcamera. Assume that two spots (
P
a
,
P
b
) are produced in a CCD image as a result of the laser projection.
D
k
is thelength of the triangle base formed by the ﬁeld of view of the CCD camera. The pixel count in terms of externalclock pulses between the projected spots (
P
a
,
P
b
) is
N
k
.Because of the alignment of the laser projectors with theCCD camera, the projected spots (
P
a
,
P
b
) or (
P
c
,
P
d
)appear on the same scan line of the image signal, no matter how the CCD camera is moved or rotated. By using a circuit designed for counting external clock pulses betweenthe projected spots (
P
a
,
P
b
),
N
k
can be easily obtained.Similar derivation of
N
p
applies to the projected spots(
P
c
,
P
d
), and will not be reiterated. As shown in Fig. 1, the distance between two projectedspots in the CCD image will be the same, irrelevant to thephotographing distance between the object and the CCDcamera as long as the laser beams are projected in parallel. That is, the projected spots (
P
a
,
P
b
) and (
P
c
,
P
d
) have thesame distance
D
min
. Irrelevant to the photographing distance, the maximal pixel count in terms of external clock pulses is
N
max
depending on the clock frequency
f
c
. That is,for each valid scan line of a CCD image, the maximalexternal clock count
N
max
is the same and can be obtainedin advance.Given
D
min
and
N
max
,
D
k
can be determined via
N
k
. By triangular formula, the relationship between the measuring distance and pixel counts can be obtained as
h
s
þ
h
x
þ
H
k
h
s
þ
h
x
¼
D
k
D
min
¼
N
min
N
k
(1) The distance to be measured,
H
k
, for a distant object becomes
H
k
¼
N
max
N
k
1
h
s
þ
h
x
(2) That is, given
h
s
and
h
x
is the distance to be measured,
H
k
,can be obtained as long as pixel count
N
k
in terms of clock pulses between two projected spots in the CCD image canbe obtained. Derivation for
H
p
is similar, and is thereforeneglected.
2.2 Determination of the op for CCDcameras
To construct a measuring system suitable for all kinds of CCD cameras, the distance
h
s
between the op and thefront end of the camera needs to be established. Fig. 3shows a proposed method for obtaining an accurate
h
s
for a speciﬁc CCD camera, in which a muzzle limiting the view angle of 2
u
s
is mounted on the CCD camera. With reference to Fig. 3, when the horizontal ruler ispositioned at (A
1
, A
2
) and (B
1
, B
2
), the distance betweenthe front end of the CCD camera and (A
1
, A
2
) and (B
1
,B
2
) is
h
m1
and
h
m2
, respectively. By triangular relationship, we have
h
s
þ
h
m2
¼
12
D
m2
cot
u
s
(3)
h
s
þ
h
m1
¼
12
D
m1
cot
u
s
(4)Subtracting (3) from (4), we obtain
h
m1
h
m2
¼
12
D
m1
D
m2
cot
u
s
cot
u
s
¼
2
h
m1
h
m2
D
m1
D
m2
(5)
Figure 2
Disposition of Laser A, Laser B and op of the CCDIET Sci. Meas. Technol.
, 2009, Vol. 3, Iss. 3, pp. 197–207 199doi:10.1049/iet-smt:20080055
&
The Institution of Engineering and Technology 2009
www.ietdl.org
Authorized licensed use limited to: National Central University. Downloaded on November 25, 2009 at 08:47 from IEEE Xplore. Restrictions apply.
Alternatively, we can obtain
h
s
þ
h
m2
h
s
þ
h
m1
¼
D
m2
D
m1
by dividing (3) by (4). Thus, the intrinsic parameter of thedistance between the op and front end of the CCD camera for a speciﬁc CCD can be obtained as
h
s
¼
h
m1
D
m2
h
m2
D
m1
D
m1
D
m2
(6)
2.3 Alternative expression to determinethe distance to be measured
The mechanism in Fig. 3 has provided a feasible way indetermining an accurate
h
s
for any CCD camera. Becausecot
u
s
is available, the formula for determining the distanceto be measured can be rewritten as
H
k
¼
12
D
k
cot
u
s
(
h
s
þ
h
x
)
¼
12
N
max
N
k
D
min
cot
u
s
(
h
s
þ
h
x
) (7)Note that the op may not be the same for different CCDcameras. By using the mechanism shown in Fig. 3, anaccurate
h
s
for a speciﬁc CCD camera can be accordingly determined, which implies that any CCD camera can beused for measuring distance via the proposed approach.In summary, the proposed measuring method considersthe adjustment of
h
x
, overcoming the drawback inmeasuring distance for objects situated in the deadzoneusing existing IBDMS approaches [16–21]. Theestablishment of the intrinsic parameters
h
s
and cot
u
s
for any CCD cameras also provides more accurate results for the measuring system.
3 Identiﬁcation of the projectedspots
To achieve a satisfactory result of the measurement, accurateidentiﬁcation of the projected spots in a CCD image isessential. In this section, disposition of the laser projectorsand circuits to identify the green projected spots from a video signal for obtaining the clock count
N
k
between theprojected spots are discussed in more detail. Concernsabout the diffusion effect of projected spots in images arealso addressed in this section.
3.1 Disposition of the laser projectors
As mentioned earlier, it is essential to position Laser A, Laser B and op of the CCD camera [16–20] so that two laser beams are formed in parallel with the optical axis of the
Figure 3
Mechanism for obtaining an accurate h
s
for a CCD camera
200
IET Sci. Meas. Technol.
, 2009, Vol. 3, Iss. 3, pp. 197–207
&
The Institution of Engineering and Technology 2009 doi:10.1049/iet-smt:20080055
www.ietdl.org
Authorized licensed use limited to: National Central University. Downloaded on November 25, 2009 at 08:47 from IEEE Xplore. Restrictions apply.

Search

Similar documents

Tags

Related Search

Prediction of Garment Drapability Based on FaEFFECTS OF TEACHING MATERIAL BASED ON 5E MODEWorks Based On The Hunchback Of Notre DameThree-dimensional computer modelling of RomanTheory and History of Three dimensional ImageDevelopment of electrodes based on conductiveThree Dimensional Space (Mathematics)Plays Based On European Myths And LegendsMusicals Based On WorksMusic Based On The Bible

We Need Your Support

Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks