3D DIGITIZING FREE FORM SURFACES BY OPTICALTRIANGULATION LASER SCANNING
Ph.D. RadomirVUKASOJEVIĆ,M.Sc.ŽeljkoRAIČEVIĆ, M.Sc.SimoŠALETIĆUniversity of Montenegro Faculty of Mechanical Engineering, Cetinjska br.2, Podgorica, Crna Gora
,Abstract: The paper provides an overview of the developed system of digitization and determines their suitability to interact with other subsystems of the computer integrated manufacturing.
The developed systems for digitization which are based on modern semiconductor laser, CCD camera and the original algorithms for detection, correction and transformation of points from the surface of the object, creating a cloud of surface points and exchange data with the software package Pro / ENGINEER are presented in the paper.
The success of the developed system is given over digitalization of more complex shapes objects.
Key words: 3D digitizing,opticaltriangulation, 3D laser scanning
1
- INTRODUCTION
Commercially is available a wide range of 3D digitizing technology based on a variety of applications that allow processing of objects in a large range of sizes. Often used terms: 3D scanning, laser digitizing and digitalshapesampling and processing forms.[1]
Applications of surfaces generation from data points captured from real objects, by optical 3D scanners and other technologies include reverse engineering, quality assurance in industry, medicine, art,, metrology, product design etc.
Data obtained from 3D digitizing technologies are often far from being perfect, and results are usually point clouds.
One scanningprocess will not lead to a complete model for most of objects. It is necessary several scanning processes from different directions in order to obtain information from all sides of the objects. Scan must be brought into a common reference system, and thencreate a complete model. [2]
Reconstruction of the shape of object, often, is done by processing a wide range of data obtained by laser scanning. Algorithms processed cloud points trough a polygonal approximation mesh. Procedures that enable the direct optimization and evaluation of the scanned data is accelerated and improved. There are more surface meshing concepts: Marching-Cubes, Delaunay triangulation/tetrahedrization, and the fitting of parametric surfaces (e. g., B-Splines or B´ezier representations) etc.
3D digitizing methods can be classified into passive and active. Passive methods, in most cases, do not require special hardware. The accuracy of these methods depends mainly on the resolution of the system for imaging and possibility to identify shapes in the picture. Passive 3D technology generally uses one of three techniques of measuring length: measurement based on photogrammetric, stereoscopically and techniques of reconstruction. [3]
Active methods have two approaches: contact and noncontact. Contact method is characterized by a contact object and the sensor. Noncontact methods, in general, are based on the reflection of radiation or light and recording reflection or transmission of energy. Reflection methods can be optical and non-optical, and transmission methods, are mainly based on X-rays.
3D laser scanning is an active, non-contact, non-destructive technology that captures the shape of physical objects using laser light. Laser scanning system works on principles of acquisition data of points or generation 3D point cloud from the surface of the object. For most 3D applications point clouds are not directly acceptable, and they have to be converted to triangle mesh models, NURBS surface models, or CAD models. Recording and measuring the spatial structure, in general, can be achieved by: triangulation, measurement based on time and phase shift. There are and other hybrid technology or combination of technologies for scanning, three-dimensional linear laser interferometry, Conoscopic holography and so on.
Non-optical methods works on the principle of determining distances by measuring the time needed to get light to the object and to return. Optical methods obtain shape by measuring the reflection of light projected on the object. Optical methods are not suitable for translucent surfaces.[4]
- DATA PROCESSING
In analyzing data obtained by laser scanning is usually necessary to be processed millions of points. Processing point cloud of real objectsgenerated with 3D digitalization process contains a certain level of errors. Reconstruction based on raw, data points will result in inadequate CAD model, which degrades the accuracy of the reconstruction of object surface.
Depending on the applied 3D digitalization technology there are various problems with results derived from the measurement procedure, the geometric complexity of the object, purpose of models etc. Regardless of 3D digitalization technique appear errors related to the point cloud, such as the presence of noise, outliers, gaps, large number of data points, disorganization and a lack of data.[5]
The presence of noise, points out of rangeetc. requires the elimination of errors and regulates the data points. Pre-processing enables filtering of errors,arranging and reduction of data points.
- LASER TRIANGULATION
Noncontact systems process geometric 3D data from complex surfaces. It is possible to obtain commercial devices based on wide variety of 3D optical sensing techniques, such as: laser scanning triangulation (3D Scanners, Cyberware, Digibotics, Laser Design, Vitronic); Moiré Fringe Contouring (Wicks and Wilson, Breuckmann, InSpeck); Phase Measuring Profilometry ([TC]) Digital Stereo Photogrammetry (TCTi).
Laser triangulation is a reliable noncontact technology for digitizing surface. Development of CCD optical sensors appliance of laser triangulation technique increased significantly.[7]Laser triangulation is an active stereoscopic technique where the distance from the object is calculated by the directional light source and video camera. The laser beam is deflected from a mirror on the scanned object. The object scatters the light, which is then collected by a video camera located at a known distance from the laser. Using trigonometry, calculate the 3D space coordinates (X, Y and Z) of the points, figure 1.
Fig. 1. 3D laser triangulation
Since the illuminated area on the optical sensor sees a group of pixels with different intensity of light, it is necessary with specific functions determine the position of points on the picture. From algorithm for determining the position of the point, depends the accuracy of digitization.[6]
Scaning method determines the relative motion of objects and light sources. It is possible that an object do not move, and that the laser-camera system, or just the laser, moving transversely, or that the object and light source are stationary and movable mirrors control direction of light which is registered by the optical sensor.
Fig.2.Types of laser light for scanning objects [6]
The light emitted by lasers and falling on the object may have a different geometric form like point, line, set of lines or points, figure 2.[6] Beam of light outlines the mark on the objectwhat making it possible to obtain information about the distance of reference points. For the measuring of complete surfaces, the projection of single points is very time consuming. If light on the object is in the form of line it is possible to determine distance of more points in a single recording, which speeds up and simplify the process, figure 3.
Fig. 3. Object illuminated by light in the form of line
Using lightening forms with a set of lines or dots increases the efficiency of data acquisition. In these three ways of lightening the optical sensor must be in form of CCD matrix.
The movement of the light source and/or optical sensor can be with constant speed or periodically. When moving is at a constant speed optical sensor detects the position of the light area on the object in fixed time intervals. In periodic motion sensor detects the position of lightening area in moment when the light source is not moving.
Scanner software calculates the value of a series of depth, which can be converted to 3D position of points in the scanner coordinate system, using a calibrated position and orientation of light sources and sensors. [7]
Standard methods for data retrieval based on optical triangulation scanning are suitable only for planar objects of uniform reflectance. Curved, unconnected and surfaces with different reflections cause a systematic distortion of the data range. Analysis of space-time can be corrected by each of the objects and achieve much higher accuracy using existing technology. When coherent light is used, such as lasers, however, appear certain places, stains, which represent the limit for the accuracy of triangulation.
3.1.CAD modeling
From digitized point cloud, obtained by measurement techniques, appropriate algorithm creates surfaces that define border of the scanned object and from topological interior creates a solid model. Modeling objects, in computer graphics, meanscreation of objects (models) using computers that is, further, may be used for: making constructive documentation, various graphic designs, technological preparation of production (CAP / CAPP / CAM) and engineering analysis.
The process of creating geometrical model consists of: defining the basic geometry of objects and its storage in database. Modeling is process of selection of basic geometric primitives and their primitives, and with connection and composing,by 2D and 3D operations model is formed.
The CAD modeling proceduredefines: geometric, functional, material, aesthetic and other properties of objects. Computer graphics uses three types of geometric models: wire frame, surface and volume.
Wireframe modelpresent a set of vertex points and border edges that define the basic characteristic of an object: contour, orientation in relation to the environment and function. Surface model is described by the vertex points, border edges and surface, which enables to define the visibility of graphical entities, creation of clear technical documentation and assemblies. Volume model is described by the apex points, border edges, and adjacent areas of the affected their capacity-solid model.
Basis of presentation of computer graphics, which gives high accuracy model, is approximation of real curved surface by Bézier's surface. Views of geometric models derived approximations of curved surfaces with flat polygons, called Facet models.
- INTERACTION OF 3D DIGITIZING SYSTEM WITH CIM
The concept of obtaining 3D data depends of technical capabilities and available equipment, which determines the technology of digitization; individual components and corresponding algorithms in the implementation.
Faculty of Mechanical Engineering from Podgorica develops non-contact digitizing systems. The 3D Center on the Faculty, through the "Machine Vision in Automation and Quality Control", for characteristic objects, develops specific software to define the models, and after that their production.
As a source of coherent light for scanning is used semiconductor laser, laser diode K52-267.manufacturers EDMUND Scientifics, with characteristics: power broadcasting 2.5 mW, wavelength 760.1 nm, beam diameter 0.7 ÷ 1.4 mm; scattering angle 60°, dimensions Ø11x62mm. On the laser diode is placed generator of line which inlet beam of laser light transfer in the plane at specific angle.
Line generator allows that light falls on the scanned object in the form of line as geometric shapes. Generally the body of laser diode allows putting in the special stand, which can be moving in all three coordinate directions.
For detection of the reflected laser light from the object is used CCD sensor R3GB CFA in digital camera TMC 9700, manufacturers PULNIX AMERICA. The camera is powered by electricity with 110/12V AC adapter, connected to the transformer 220/110V, type MA4150, manufacturer NeBoElectronics.
Camera features: size CCD 8.9x6.7 mm, resolution CCD 768x484 pixels; dimension cell 11.6x13.6 μm. The camera lens is a COMPUTAR G13-85, manufacturer Sigma.
The camera is placed on a stand that allows precise positioning in relation on the workpiece and the laser diode.
As a rotating standfor workpieces is used rotating worktable VMR-320 of universal milling FGU-32, manufacturer MZT Skopje, dimension Ø320mm. By adding a special aluminum strips with the appropriate measures increases positioning accuracy 20 times.
Processing is done on HMC 500/40Horizontal Machining Center.
Laser is set so that the emitted light line passes through the vertical axis of the rotating base, and camera is positioned from the workpiece at a distance depending on the desired resolution of imaging.
Computer for data acquisition and data processing Compaq Deskpro, with Frame Grabbergraphics card ULTRA II RGB, manufacturer Coreco Ltd., connected with the camera over the PCI interface. Image acquisition application is TCi Professional, manufacturer Coreco Ltd. Applicationallows online display and freeze image from camera.
Effective image analysis andfiltering data was performed by rejecting pixels with intensity less than preset level, i.e. setting the intensity of light pixels to zero for all pixels with intensity lower then P=200.
Convex surface of the object can cover reflected laser light from the camera lens, it leads to gaps in the images and the absence of digitization.
Light rays passing through the camera lens leads to their unequal refraction on spherical boundary surfaces, which results in deformation of the image. Only the light beam that matches with the optical axis is not distort. In addition, the reason for the deformation can be and construction of CCD sensors. Cells in CCDhave rectangular shape. Therefore, it appears compressing images in the direction of smaller size CCD matrix cells, with the coefficient equal one to another dimension of the cells.
Software that provides detection, correction and transformation of coordinates for the purposes of digitalization is Pro/ENGINEER with Pro/SURFACE, Pro/SCAN-TOOLS, Pro/MFG and Pro/NC-POST MODULS, and specific application, developed in 3D Centre Faculty of Mechanical Engineering. This application allows analyzing pictures, forming shapes of surface points, correction, filtration and optimization of data to be used in Pro/ENGINEER software.
Correction of digitized coordinates of points is the inverse process of forming. Deformation covers all pixels, from pictures.But, essential to process are only points that represent the coordinates of the points of structure, which shortens the process of correction.
Coefficient of image correction is the process of determining the validation by which the deformed image pixel transform into pixel non-deformed image. Deformation may occur due to eccentricity axis of optical lens to the center of the CCD sensor.
Images of scanned object with flat surfaces, is shown on figure. 4, and cylindrical surface figure 5. Image processing was performed through a specific algorithm steps: detection of point coordinates, interpolation and correction coordinates of points and optimization of the number of points.
Fig. 4 Visualization of the reconstructed flat surface
Fig. 5 Visualization of reconstructed cylindrical surface
Transformation of coordinates is a process that position of points in a polar-cylindrical coordinate system translated into a rectangular coordinate system.
Reconstruction Program Pro/SCANTOOL as input accepts files created in formats IGES and VDA. Post processing program allows transfer to the control unit of operating the machine.
Fig. 6. Processed segment of surface fuel unit
For processing on HMC 500/40HORIZONTAL Machining Centerusing three sequences: Profile Milling processing by cylindrical cutter; Conventional Surface -rough surface reconstruction and Conventional Surface -
finish surface reconstruction.
- CONCLUSION
Digitization is used for systematic and accurate collection and processing of data. 3D model represents group of data used to showan object in the 3D virtual space. 3D modeling means the use of polygons or NURBS. Product of 3D laser scanning and photogrammetric scanning is 3D measurable model in the form of cloud points in which every point has a spatial coordinate. For digitalization may be used commercial devices, available equipment and software packages with appropriate customization which guarantee the accuracy of the data collected.
REFERENCES
[1] Lee K.H., Woo H., Suk T. (2001) Data Reduction Methods for Reverse Engineering, The International Journal of Advanced Manufacturing Technology, Vol.17, Springer-Verlag London Limited,
[2] F. Bernardini and H. Rushmeier (2002)The 3D Model Acquisition Pipeline, IBM Thomas J. Watson Research Center, Yorktown Heights, New York, USA, Volume 21
[3] C. Teutsch (2007)Model-based Analysis and Evaluation of Point Sets from Optical 3D Laser Scanners, volume 1, Shaker Verlag
[4] Schwenke, H. et al. (2002.)Optical Methods for Dimensional Metrology in Production Engineering, Annals of the CIRP, 51/2
[5] Huang M.-C., Tai C.-C. (2000)The Pre-Processing of Data Points for Curve Fitting in Reverse Engineering, The International Journal of Advanced Manufacturing Technology 16:635-642, Springer-Verlag London Limited
[6] Otto K., Wood K. (2001)Product evaluation: A reverse engineering and the redesign methodology, Reserch in Engineering Design, Springer, Verlag, London, 1998.
[7] S. Zhang, P. Huang (2006) High-resolution, real-time 3-D shape measurement,Optical Engineering
1