In photogrammetry, an approach of automatic detection and recognition on reference points have been proposed to meet the requirements on detection and matching of reference points. The reference points used here are the CCT(circular coded target), which compose of two parts: the round target point in central region and the circular encoding band in surrounding region. Firstly, the contours of image are extracted, after that noises and disturbances of the image are filtered out by means of a series of criteria, such as the area of the contours, the correlation coefficient between two regions of contours etc. Secondly, the cubic spline interpolation is adopted to process the central contour region of the CCT. The contours of the interpolated image are extracted again, then the least square ellipse fitting is performed to calculate the center coordinates of the CCT. Finally, the encoded value is obtained by the angle information from the circular encoding band of the CCT. From the experiment results, the location precision of the CCT can be achieved to sub-pixel level of the algorithm presented. Meanwhile the recognition accuracy is pretty high, even if the background of the image is complex and full of disturbances. In addition, the property of the algorithm is robust. Furthermore, the runtime of the algorithm is fast.
Nowadays, computer vision has been wildly used in our daily life. In order to get some reliable information, camera
calibration can not be neglected. Traditional camera calibration cannot be used in reality due to the fact that
we cannot find the accurate coordinate information of the referenced control points. In this article, we present a
camera calibration algorithm which can determine the intrinsic parameters both with the extrinsic parameters.
The algorithm is based on the parallel lines in photos which can be commonly find in the real life photos. That
is we can first get the intrinsic parameters as well as the extrinsic parameters though the information picked
from the photos we take from the normal life. More detail, we use two pairs of the parallel lines to compute
the vanishing points, specially if these parallel lines are perpendicular, which means these two vanishing points
are conjugate with each other, we can use some views (at least 5 views) to determine the image of the absolute
conic(IAC). Then, we can easily get the intrinsic parameters by doing cholesky factorization on the matrix of
IAC.As we all know, when connect the vanishing point with the camera optical center, we can get a line which is
parallel with the original lines in the scene plane. According to this, we can get the extrinsic parameters R and
T. Both the simulation and the experiment results meets our expectations.
With the camera internal parameters known, to calculate the external parameters is to solve a set of highly nonlinear over-determined equations. In this paper, an improved hybrid genetic algorithm is adopted to obtain external parameters. It combines the advantages of genetic algorithm and Newton method, making it possible to obtain results with high accuracy and a faster convergence.
To determine the relative pose between an object and a single camera using correspondences between 3D feature points of the object and their corresponding 2D projections in the image, this paper proposes a target’s pose measurement algorithm based on the bundle adjustment method. This iterative algorithm can be divided into two steps: first, a reliable initial pose is computed by using only three non-collinear points; second, the optimal rotation and translation matrix of the target is estimated based on the bundle adjustment method. Experiments on simulated data and real data show that this method can get high precision with its rotation angles error within 20 arc-second and the translation error within 20 micron in ideal occasions.
The paper proposed a large field of view (FOV) measurement with calibrating the camera and measuring simultaneously.
In the measurement, the whole FOV was divided into several smaller ones with overlapping areas between each other.
The overlapping areas should contain at least 4 noncollinear feature points in each for computing external parameters
and at least 4 noncollinear control points in one of them to start the calculation. To obtain the measurement of the whole
large FOV, 2 images (or more) of each small fields of view needed to be taken from different angles. In the process of
calculation, theoretical values of the camera were used as the initial values of the internal parameters and the initial
values of external values were obtained from a new solution for P4P problem. So, the internal parameters of the camera,
the external parameters for each image, and the 3D coordinates of the feature points in the large field of view could be
acquired by adjustment method. In our experiment, the large field of view range was 500mm×500mm, the smaller ones
corresponding to each image was 200mm×200mm, and the ultimate measurement accuracy was 12μm.
The method of 3D reconstruction from multi-view of the object with single camera is proposed in this paper. The images
of the planar pattern from different views are captured by the single camera. The internal and external parameters of the
camera are calculated precisely with Two-stage camera calibration method. On this basis, the image points of calibration
pattern can be matched and reconstructed. Experimental results show that: the flatness of reconstructed planar pattern is
about 0.006mm; the position error of sign is about 0.012mm. The proposed method is high-accuracy, simple, flexible and
effective.
During the experimental process of the spatial point's position detection, we analyze the advantages
and disadvantages of the classical method, and propose an improved method. First, we interpolate the
point's data to increase the size of the image. Then use the Zernike moment to detect the edge of the
point. Finally, we obtain a coordinate of the center of the point by using the ellipse fitting algorithm,
and take this coordinate as the position of the spatial point. Experimental results in laboratory show
that, the proposed detection method on sub-pixel level is realized to detect the spatial point's position
with high-accuracy. And experimental results of relative measurement with 100 meters outdoor show
that, the repeatable accuracy of this method can reach 0.12 pixel during the day and 0.05 pixel at night.
A new method of attitude measurement of the calibration target based on machine vision is proposed in this paper.
Firstly, the internal parameters and external parameters of the CCD camera are calibrated by using the feature points on
the calibration target. Secondly, shoot the calibration target in different positions and gather any different image
information. Finally, measure the calibration target's attitude through reconstructing the positions of those feature points
which on the calibration target. Experiment results show that the standard deviation of the attitude measurement errors is
less than 10 arc-second. This method is an effective high-precision space attitude measurement algorithm which can
meet the engineering requirements.
Camera calibration is the important step for the stereo vision measurement system. The accuracy of the calibration
affects the accuracy of the vision measurement system. The calibration is realized with a planar pattern. The pattern
has some spot array with high precision. Stereo vision measurement system captures the images of the calibration
pattern at several different orientations. The parameter model is modified as the polynomial model to describe the
lens distortion. The calibration experiment demonstrates that the error of 3-D reconstruction is 0.020mm.
This paper discusses an extension of image mosaics technology of the 2D coordinates measurement technique.
Fourier-Mellin Transform (FMT) is widely used as an image mosaics method. Based on analyses of frequency
spectrum, FMT automatically stitches images with relationships of translation, rotation, scaling. The course of FMT is
simulated and optimized, and the character of FMT for shifted images is educed. Furthermore, the particular rules for
polar coordinate transform of rotated and zoomed images are advanced, making noise depressed and the feasibility of
the algorithm greatly increased. Moreover, The proposed method that adopt filter technique aiming at finite discrete
images can eliminate efficiently the artifact in the Fourier-Mellin space, and obtain significant correlated peak values
and resist the influence of noise interference. Finally, making use of the separation pattern technology based on
difference calibrations to evaluate the precision of registration image. Experimental results demonstrate this method is
effective.
KEYWORDS: Reflectors, Cameras, Photogrammetry, Spherical lenses, Telescopes, Calibration, Retroreflectors, CCD cameras, Radio telescopes, Control systems
China has embarked on a project to build the world's largest radio telescope, the Five hundred meter Aperture
Spherical Telescope (FAST), in a karst depression in southwest Guizhou Province. The telescope is of a
modified Arecibo type. We suggest an active main reflector that is spherical in the neutral state but the
illuminated aperture of 300m in diameter would be adjusted into a proper paraboloid such that a simple feed
could be used at its focus, since the "spherical correction" has be done on the ground. The feed cabin at the
focus is supported and driven directly by cables controlled by computer, which avoids a heavy and expensive
feed supporting system. Newly developed method and technology for determining the spatial position of 2400
nodes on the main reflector of the FAST are introduced in this paper. Base on the measurements of the position
of node under which the down cable are linked, a loop feedback control enables accurately driving the spherical
reflector deformed to paraboloid. The key technique of this implementation is the precise measurement of
2400 nodes. In this article, we introduce the scheme of simplified photogrammetry aiming at no lateral shift
situation; we analyse the influence of lateral shift; and considering lateral shift, we propose a scheme using
rotation platform plus double-eye camera to accomplish the dynamic measurement of the reflector. The result
of analysis and testing shows the feasibility and effectivity of the scheme of measurement.
The aims of LAMOST(Large Area Multi-Object Fiber Spectroscope Telescope) optical fibers positioning system is carrying out 4000 fibers minutely position quickly on the focal plane plate. Base on the dividing domain, we are putting forward parallel controllable optical fiber positioning system, this system consists of several parts as follows: In the focal plate of LAMOST, A aluminous alloy plate with plate diameter 1.75 m, globe radius is 20m. Over 4000 holes are bored on the focal plate; one optical fiber positioning unit of double revolving freedom device is inserted in each holes of focal plate, it is drived by two micro-stepping motor and positioning one fiber-end, focal plate is sustained by 8 steel tubes on the focal mechanical framework; for driving 8000 stepping motors, a control system is needed; and a measuring system with 4K surface CCD is used to calibrate the fiber's position, besides a few accessorial devices for example 4000 wire and fiber setting up need to plan elaborately, According to plan, parallel controllable fiber positioning system will be made in the next three years.
Such as in LAMOST (Large Sky Area Multi-Object Fiber Spectroscopy Telescope), many photometric measurement systems need to reach sub-pixel accuracy with area scan CCD camera. The separation patterns are used to calibrate a single-camera with high precision. Several separation calibration patterns with small size are put on the position of object plane of the camera. Each pattern has some spot array with high precision. The position of each reference point on the image plane of the camera is calculated. The coordinates of the reference points on the calibration patterns are used to calibrate the camera. The curved-surface fitting method is applied to fit the perspective relationship between the object plane and the image plane. The integer pattern with large dimension can be replaced by the several small differential patterns in the situation of large field. The difficulty to manufacture the large pattern is avoided. The experimental results show that the mean value of residual error is less than 0.002mm with the separation calibration method.
This paper proposes research of a structure of LAMOST fiber positioning medium-term system; its construction and its tests are briefly introduced. This medium-term system includes several parts as follow: a main control computer, a unit controller, a set of drive circuits, 19 optical positioning units that positioning 19 optical fiber ends on the small simulate focal plane with diameter of 180 mm, a CCD camera, a frame grabber, and control programs. Tests on the system have indicated that positioning precision of 19 units is less than 0.04 mm on the whole focal plane with diameter of 180 mm. On medium-term system, some important problem for LAMOST building has test and research, for example: Fiber positioning precision, mechanism interference among the units, anti-jamming of drive circuits, unit's work life-span and reliability, temperature raising, etc. Test results have established stability foundation for LAMOST construction.
The paper proposes the measurement system for the fiber positioning unit of LAMOST(Large Sky Area of Multi-Object Fiber Spectroscope Telescope). It consists of an area CCD sensor, an image acquisition card, and a lens. The fiber is illuminated by light source from one end. The end of the fiber on the focal plate is imaged on the area CCD sensor by the lens. The image of the fiber end is acquired by the area CCD sensor, and transferred into a computer by the image grabber. Some pro-processed methods are used to process the digital image of the fiber. According to further digital image processing, the position of the fiber is obtained. The paper focuses on the calibration method of the digital area CCD camera. The measurement system calibrates the camera with the calibration board. The calibration board has some holes illuminated by an area LED. The positions of those holes are pre-measured precisely. Then, the systematic error of the measurement is figured out through the calibration procession. The optical aberration is fitted by the quartic surface. The measurement system can measure the position of the fiber on the positioning unit precisely. The precision of the measurement system is 0.010mm.
The architecture of the software which controls the LAMOST fiber positioning sub-system is described. The software is composed of two parts as follows: a main control program in a computer and a unit controller program in a MCS51 single chip microcomputer ROM. And the function of the software includes: Client/Server model establishment, observation planning, collision handling, data transmission, pulse generation, CCD control, image capture and processing, and data analysis etc. Particular attention is paid to the ways in which different parts of the software can communicate. Also software techniques for multi threads, SOCKET programming, Microsoft Windows message response, and serial communications are discussed.
KEYWORDS: Control systems, Control systems design, Structured optical fibers, Computing systems, Stars, Data conversion, Capacitors, Sensors, Position sensors, Optical fibers
This paper describes the design of control system of fiber positioning system. The fiber positioning system has more than 4000 fiber units with 2 stopping motor and 2 start position sensor in each unit, and whole units will be assembled at 1.75 meter diameter focal surface of LAMOST, the mechanism and control system have demanding requirements for high precision position control. Detail design, testing and performance evaluation is described in this paper, a special control unit which can control and monitor more than 20 fiber with 1000Hz driving frequency of stepping motor and response start position sensor with one pulse of stepping motor in each fiber unit is set as a CAN bus node, 200 control units combine a can bus real time control system which can control the whole fibers move to new position in 3 minutes. In order to get high precision position in this open loop control unit, a very simple and small sensor is used to eliminate the accumulate errors of mechanism with resetting the start position, and compensation data is measured and set in control software to diminish the mechanical transmission errors. For testing the mechanism and control system, a small fiber positioning system with 19 units have been made.
This paper proposes an optical fiber positioning unit device for LAMOST(Large Area Multi-Object Fiber Spectroscope Telescope), It consists of a central shaft revolving mechanism, and eccentric shaft revolving mechanism relative to central shaft. The central shaft turns round at the range of -180° to +180° and the eccentric shaft turns round at the range of -90° to +90° driving by each control motor. When positioning, the optical fiber end moves on the focal plate throughout, and can never deviate from focal plane. It has simple structure, could be machined and assembled and taken down easily and could be ensured machining practices easily, so could be reduced manufacture costs. The unit sets mechanical electrical zero position detecting device to reduce the accumulate error of multi-positioning. Testing result have demonstrated this new double revolving optical fiber positioning unit device can accord with the demand of LAMOST entirely.
In present paper we have proposed an optical fiber positioning system for LAMOST (Large Sky Area Multi-Object Fiber Spectroscopy Telescope). In this system, the convex focal plate of the telescope (Its diameter is 1.75 m) is divided into about 4000 individual domains. Each domain contains a controllable unit which is named X-unit. Each unit is specified with the position of its domain with polar coordinates (the angle and the radius distance from the center) by using its center point as reference. Each X-unit can be moved smoothly by verifying the two parameters of polar coordinates. An optical fiber is hold with X-unit and introduced to spectroscope, thus the optical fiber can be moved anywhere within its designated circular domain at the convex focal surface. In addition, the circular domain are overlapped in a `honeycomb-like' arrangement to ensure that there are no blind spots on the convex focal surface, such as that the radius distance of the units is 25 mm, and the positioning range of the optical fibers is 30 mm, thus there will be no `blind areas'. The structural parameters of the X-unit could also be optimized to limit the directional error of the axis in a certain range. Each unit is driven by two stepping motors and mobilized by computer in a way to ensure that the units do not collide with each other. It is possible that very high or very low star densities in areas of the field of view may reduce the observational efficiency of the telescope. This should not a significant problem by using optimization of the observation program.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.