With the new reorganization of the working groups within Commission V, there are significant interest in digital photogrammetry by all working groups within the Commission. The number of papers devoted to this subject in this Zurich Symposium is a clear evidence that computer vision techniques will be a major force in the automation of close-range photogrammetry in many traditional as well as new application areas. Within the new structure of the working groups, WG V/1 will have responsibility in areas related to the development and implementation of fully automated and real-time systems. More specifically, the terms of reference are as follows: 1. real-time vision systems for metric measurement; 2. near-real-time, but fully automated, vision systems with relaxed time constraints; 3. system hardware and software integration; and 4. demonstration of real-time and near-real-time systems in actual application environments.
An International Robomation/Intelligence (IRI) DX/VR Vision System and a PC/AT vision system are being used to conduct research on the application of vision systems for performing metric measurements in engineering construction and manufacturing. The IRI DX/VR system is a UNIX-based, multi-user program development workstation with a host computer that is capable of performing about 1 million instructions per second. The PC/AT system employs a Zenith 386 personal computer, two frame grabber boards to permit simultaneous imaging from two cameras, and MS-DOS operating system. Both systems are equipped with 512x512 CCD cameras. An expanding package of software is being developed to perform, on either vision system, many photogrammetric functions, including: camera calibration, image correlation, and feature matching. While the IRI system provides more speed in performing basic image processing operations, the PC system has the major advantage that a large library of commercial software is available for performing data analysis. Geometric distortion characteristics of both line-transfer and frame -transfer CCD cameras are being studied.
The Canadian contribution to the International Space Station "Freedom" is the Mobile Servicing System (MSS). The MSS (Figure 1.0) is a collection of robotic elements configured to support the assembly, maintenance and servicing of the space station. The major components of the MSS are: Ã‚Â° Space Station Remote Manipulator System (SSRMS): a large robotic manipulator arm with seven degrees of freedom used primarily for handling large objects on station including the capture and docking of the shuttle orbiter. Ã‚Â° Special purpose Dextrous Manipulator(SPDM): a smaller two arm robotic device capable of performing robotic servicing tasks requiring dexterity. Ã‚Â° Mobile Servicing Centre (MSC): a mobile base for the MSS from which the robotic manipulators can be operated remotely and used to transport payloads around station. Although these manipulators will usually be controlled directly or remotely by astronauts, a fundamental component of the MSS Control Equipment (MCE) is an Artificial Vision Unit (AVU), which will play a major role in both manual and automatic control of the manipulators and cameras. The baseline specification states that the AVU will support the rendezvous and proximity operations including payload tracking, payload capture and payload berthing3. In one mode, the AVU shall provide data to the astronaut operator in the form of textual and graphical displays. In another mode, the AVU data shall allow the closed loop control of MSS manipulators for these operations. The baseline also requires Fine Robotic Service tasks to be provided with machine vision support. This paper deals with some aspects of a program to develop the necessary machine vision technologies.
The US Federal Highway Administration and the Center for Mapping of The Ohio State University are developing a prototype mapping vehicle that integrates a stereo-vision system, a GPS receiver, and some other sensors, to automatically collect data of the environment of roads at highway speeds and store this information in a GIS. Thirty eight states and one Canadian province are sponsoring this effort together with NASA. In this paper we discuss the integration of the stereo-vision system with a GPS-receiver and an inertial system in a mobile workstation. The major task of the vision system is the acquisition and storage of digital image pairs once a second. The shutters of the camera are synchronized by the clock of the GPS receiver. The system will be calibrated by analytical methods using a testfield of control points. The offset between the GPS antenna and the vision system is determined by a rigorous bundle- and geodetic adjustment. Beside the storage of images the vision system is used for sequential triangulation of stereo pairs along the road, to bridge over areas without satellite signals. We also try to extract the edges of roads automatically and compute its width. As the construction of the prototype is presently ongoing, preliminary results and the configuration of the hardware are presehted.
Medical and industrial needs exist to track the position, volume and geometry of moving objects whose shapes change. Video imaging offers frame rates of sufficient frequency to support real-time sensing of fairly rapid changes in the real world. It is the processing of the video frames which represents the bottleneck in a "real- time" implementation. However, under certain limiting constraints, namely near-symmetry around a curved axis, real-time measure- ments are feasible with standard PC-based software to obtain 3-D positions, shape and volume of a moving and changing 3-D object. The paper describes a novel stereo-photogrammetric analysis system based on object silhouettes. Implementation involves the end-to-end system from video cameras to user interface and reporting of results. A medical application's scenario puts high demands on robustness and simplicity of the user interface, while operating in an entirely automated fashion. "Real-time" represents the need to process an image pair each 2 seconds. Accuracies of Ã‚Â±1% in dimension and 3% in volume are required and permit one to accept certain shortcuts in the photogrammetric approach.
Since 1975 lateral-effect photodiodes have been well-known opto-electronic sensors for the measurement of imaged light sources. They provide a simple evaluation of the image coordinates of the centroid of an imaged light source and have excellent dynamic characteristics. Due to the physical principle of these sensors the accuracy of 3D-measurement is greatly influenced by reflections and distortion causing large non-linearities. This paper describes investigations concernig the theoretical and experimental accuracy behaviour of a developed photogrammetric system equipped with lateral-effect photodiodes.
The new Working Group in Architecture and Archaeology was established as part of Commission V in 1988. However, this was not the start of Commission V involvement with Architectural and Archaeological Photogrammetry, as for a number of years we have had an "Associated" group.
Before the restoration of works of art we need to execute metric, qualitative and storic researches to acquire a deep knowledge of the object on which the opera- tion will be executed. We intend to expound and explain. the problems concerning metric surveys executed with the stereophotogrammetric method on particular "movable" works of art. These problems include: the acquisition of control points with the Wild-Leitz MiniRMS2000 system, the selection and definition of the essential lines of the stereoscopic mo- del, the transfer of acquired data to a CAD system, the processing to these data for the creation of thematic drawings or parallel views or for the separation of the object in parts, etc. We will produce surveys and processing of a painting on wood of the Maestro di S. Francesco (XIII century, m 4.90 x 3.54 x 0.15) and a polyptych of S.Antonio on a board and a frame also on wood painted by Piero della Francesca (XV century, m 4.10 x 2.00 x 0.12). Both of them are of the Soprintendenza ai Beni Architettonici Artistici Ambientali e Storici dell'Umbria - National Gallery of Perugia. We will also produce parts of a painted glass window of the Orvieto Cathedral apse.
The Laboratory of Photogrammetry of the University of Architecture in Venice is involved in a research program aimed to quantitative measurement with respect to reliable and precise informations. The walled city of Cittadella (one mile of medieval walls with four gates and thirty-two towers) will shortly go under a restoration program financed by the Italian Governement, the Veneto Region and the municipality of Cittadella. The question in point is to find the most suitable system to respond to a series of problems that different disciplines will raise which can be here only briefly summarized: 1 - archeological analysis will be concerned in determining the various periods of construction of the different "layers" of the wall; 2 - statical analysis and geodetical surveys will try to determine the quality of the intervention of restoration; 3 - chemical and physical measurements will ascertain the effect of erosion and deterioration of materials testing the most convenient procedures for future protection. For all these purposes a specific field of experiment has been recently set up where photogrammetry will be tested to combine "classical" expertise with the "new" approach of computer vision, 3D models and cost-benefit analysis.
ABSTRACT PADRAS,PASCO Archaeological DRAwing System, has recently evolved into two directions. The first direction of evolution is utilization of CAD system.The previous mobile type PADRAS system consisted of photo laboratory,portable analytical plotter and X-Y plotter in a microbus.By introducing AutoCAD system into analytical plotter(PADRAS-P3D) and total station surveying system(PADRAS-T3D), we have obtained real time 3D digitizers of a CAD system applied to drawing of archaeological research.Total station surveying system has realized on-line and real-time visualization of an archaeological site on a loptop PC display within the environment of AutoCAD. The second direction is related to connection between CAD system and Data Base system.PADRAS systems attach attributes to graphical features which are compatible to relational data base software(R:BASE PRO).Aiming to digital historical maps or Atlas,PADRAS system,as a whole, evolves out of an approach of Drawing System,Relational Data Base into a construction of Geographic Information System, utilizing new combination among photogrammetry,terrain surveying,CAD system, Data Base systems and Geographic Information System,such as ARC/INFO.
In the field of Maya research archaeological studies in architecture play an important part. In the case of Santa Rosa - XTAMPAK the influences of different cultural regions have to be traced. Structures like this situated in border areas, especially require exact surveying so that signficant archaeological assertions can be derived. For the main palace of this site a threedimensional digital architectural model was established, using geodetic-photogrammetric methods. Possibilities of visualization with the help of recently developed technology are shown by examples: certain details of the palace as well as an inner staircase have been digitally recorded. This certainly means considerable help for further study.
The practical application of analytical photogrammetry permits digital data being used in CAD systems such as AutoCAD. In the case of fire damaged buildings sometimes only small or partial examples of detail remain, but by combining photogrammetry and CAD this detail can be recorded and repeated to replicate the original design. This paper looks at some actual examples of analytical photogrammetry and CAD joining hands to help reconstruct some historic buildings including a famous Wren church in the City of London.
The state-of -the-art in sensor techniques for close-range photogrammetry is characterized by two major trends. In the field of photographic sensors reseau cameras are gradually superseding conventional metric systems and leading to camera systems of very high accuracy. Fully digital imaging systems are also advancing, and particular use is made of CCD cameras with standard video output. Higher resolutions can currently be only achieved if the object is recorded sequentially by a moving CCD sensor in the image space. Additional illumination devices and object targeting have to be involved in the photogrammetric recording process. Image recording systems are only one component of a total system. In order to judge the efficiency of the system the complete object recording and data processing procedure must be taken into account.
The geometrical accuracy of modern CCD sensors is typically better than 1/100 of the pixel period. For this reason CCDs are almost ideal tools for many optical metrology applications. Traditional video framegrabbers are ill suited for these applications because framegrabbers cannot acquire the video signals synchronously with the CCD's pixel clock. A universal CCD data acquisition system was developed, capable of digitizing video signals from almost any CCD camera, up to a speed of 20 MHz with either 8 or 10 bits of accuracy. Practical problems had to be solved for applications with standard CCIR CCD video cameras (560v x 600-800h resolution, 50 fields per second), and scientific high-resolution CCDs (1024v x 1024h resolution, 10-40 frames per second). The experimentally determined accuracy of the position of an object, measured with a CCD image sensor using 10-bit pixel- synchronous digitization is better than 1/100 of the pixel period, obtained by averaging many measurements. This experimental accuracy is not as good as the theoretical calculation of the precision indicates. The main reason for this are: pixel response non-uniformity, fabrication tolerances of the CCD sensor, non-ideal sampling of the video signal, and residual timing jitter in the pixel-synchronous sampling.
A newly developed imaging principle, two dimensional microscanning with Piezo-controlled Aperture Displacement (PAD), allows for high image resolutions. The advantages of line scanners (high resolu- tion) are combined with those of CCD area sensors (high light sensitivity, geometrical accuracy and stabi- lity, easy focussing, illumination control and selection of field of view by means of TV real-time imag- ing). A custom designed sensor, optimized for small sensor element apertures and color fidelity, elimi- nates the need for color filter revolvers or mechanical shutters and guarantees good color convergence. By altering the computer controlled microscan patterns, spatial and temporal resolution become interchange- able, their product being a constant. The highest temporal resolution is TV real-time (50 fields/sec), the highest spatial resolution is 2994 x 2320 picture elements (Pels) for each of the three color channels (28 MBytes of raw image data in 8 sec). Thus for the first time it becomes possible to take 35mm slide quali- ty, still color images of natural 3D scenes by purely electronic means. Nearly "square" Pels as well as hexagonal sampling schemes are possible. Excellent geometrical accuracy and low noise is guaranteed by sensor element (Sel) synchronous analog to digital conversion within the camera head. The cameras principle of operation and the procedure to calibrate the two-dimensional piezo-mechanical motion with an accuracy of less than 0.2[tm RMSE in image space is explained. Latter uses a fast algo- rithm to estimate the shift between image pairs with an accuracy of about 1/1000 Sel, based on greyvalue gradients modelled as second order polynomials. The remaining positioning inaccuracy may be further reduced by adaptively postprocessing the high-resolution images.
The small dimensions of the focal planes of solid-state sensors and the limited resolution of the pixel array decrease the potential applications of vision systems. Zoom lenses can enhance the flexibility and capabilities of vision systems by providing the user with continuous access to a full range of focal length settings without degrading geometric accuracy. The key to the use of such lenses lies in the inherent stability of array cameras. Preliminary test results showed that when the focal length setting of a zoom lens was changed, significant changes occurred in both the position of the principal point and lens distortion characteristics. However, these changes were found to be either linear or curvilinear over the entire range of zoom. Moreover, both the magnitude and pattern of the changes were found to be highly stable over a period of weeks. Thus, the results indicated that the changing interior geometry of zoom lens camera systems can be pre-calibrated and then applied during the measurement process. This paper reports on experimental results and distortion models using two zoom lenses with focal lengths ranging from 12.5mm to 75mm.
High resolution digital cameras have become an interesting tool in digital photogrammetry. At the Technical Uniyersity of Braunschweig, Department of Photogrammetry and Image Processing, several cameras, including the Reseau Scanning Camera, Videk Megaplus and the ProgRes 3000, have been under investigation for photogrammetric purposes. The different ways to achieve high resolutions are briefly described, as well as the method of investigation, including digital point measurement and testfield triangulation. The results are presented in the last part together with practical experiences made in camera handling.
During the recent years, the use of small format cameras, whether metric or nonÃ¢â‚¬â€metric, has flourished again, no doubt mainly due to the rapid development of the personal computers. In this paper the photogrammetric applicability of these cameras is reviewed, discussed and assessed. Special emphasis is given to their implementation in archaeological or architectural photogrammetry. Finally their prospects are investigated and a bold attempt is made to forsee their future.
In this paper, an analytical photogrammetric evaluation system, developed using off- the-shelf optical, mechanical and electronical components, will be introduced. The measurement of the diapositive (up to the format 24 x 24 cm') is based on two low- priced, hut precise x, y-positioning systems, driven by step motors via spindles. Stage positioning in comparator or model mode is provided by two trackballs. The software contains a special program for 3D representation of buildings. This paper reports on the possible increase in efficiency by using high resolution photography.
Since August 1988 the University of Bamberg works with the analytical photogrammetrical evaluation system ROLLEIMETRIC MR2 in the Institut of Ancient Monuments survey. As so far in this field measured drawings had been carried out by hand, it was to be evaluated if the photogrammetry would obtain sufficient results for the building research. Besides the low-cost equipment, the system provides the advantage of the possibility to be used by non-photogrammetrists. For the building research this is of great importance. With three examples it is shown that ROLLEIMETRIC MR2 is a useful tool in the building research and is demonstrated how important the cooperative work between photogrammetrist and the user of the drawings is for obtaining best results. The advantages and disadvantages of the System are discussed.
This paper outlines research being undertaken in the Civil Engineering Department at Leeds University, England, using a new, low-cost analytical photogrammetric plotter. The project, which is funded by the UK Science and Engineering Research Council ,involves the use of an MPS-2 desktop analytical plotting system, manufactured by Adam Technology of Australia, to produce architectural drawings and to monitor crack propagation in structures. Details are given of the working system which consists of metric and non-metric cameras, a personal computer, an interface to AutoCAD software and a rolling drum plotter. The areas of application being investigated are illustrated by studies of buildings and structures in the civil engineering laboratory of the University and around the Leeds area. The problems encountered are summarised and an assessment of the system is made. The paper concludes with a discussion of the progress of the research project and reviews the work still to be undertaken.
This work deals with the application of the collocation method to the control of the floor movements of the St.Marcus' Basilica in Venice. Results and prospects are examined and some graphic representation systems are proposed which make easy the reading and the interpretation of the movement itself.
A Rollei 6006 partial-metric camera was equiped with special objectives of 50 resp. 28 mm focal length to control and measure the surfaces of small sand-stone plates (complex research programme of the Ministry of Science and Technology of the FRG). The lenses were mounted in retro-position, a surface of the stone samples of 5 cm x 5 cm resp. 2.5 cm x 2.5 cm can be photographed. The depth of focus was 8 resp. 6 mm. An optical flat grid was used for the system calibration. The measurements were carried out by a Zeiss Planicomp C100. The standard deviation of eleva- tion after bundle adjustment was 16 m for the 50 mm-lens resp. 9 m for the 28 mm-lens. The planimetry was much better. Results of bundle adjustment will be shown and a CAD-evaluation will demonstrate the stone weathering.
The department of architectural photogrammetry of the Ministry of Culture has developed a program to create synthetic images for architectural analysis from three- dimensional photogrammeteric datasets.
A complete documentation of an architectural object makes it possible to reconstruct the whole object both geometrically and pictorially. In the past the geometric documentation was performed using line maps, e.g. from photogrammetric restitution, while the pictorial aspect was mainly covered by conventional photographs. Today digital photogrammetry and digital surface description techniques allow for the combination of the geometric and the pictorial aspect in a complete documentation of an architectural object. This paper presents some experiences with visualization of architectural objects. At first, various ways of digital surface modelling are described. Then possibilities of visualizing digital surface models (DSMs) are shown in practical examples. The paper is concluded with some recommendations.
The Working Group V/3 of the International Society for Photogrammetry and Remote Sensing (ISPRS) was founded in 1988 at the 16th ISPRS International Congress. Since that time it works on " Image Analysis and Image Synthesis in Close-Range Photogrammetry ". The following report deals with concepts and topics which should deeply be investigated during the second half of the research period.
Abstract. The study of the aspect graph of a three-dimensional object has recently become an active area of research in computer vision. The aspect graph provides a complete enumeration of all the possible distinct views of an object, under a particular model for the viewpoint space and a particular definition for "distinct." This paper gives a tutorial introduction to the aspect graph, surveys the current state of the art in algorithms for automati- cally constructing the aspect graph, and describes some possible applications of aspect graphs in computer vision and computer graphics.
This paper presents parts of a research project designed to meet the need of automated or semi-automated object measuring and description in digital images. Different sources of information are used to provide the necessary information to generate different hypothesis regarding the object. Information can be provided by feature extraction, logical knowledge of the object, a CAD-model or similar. The main objective of the system is not in the field of object recognition but rather object description. The searched objct is prescribed as a generic model. To be able to compare hypothesis of models of different complexity other measures than maximum likelihood estimates have to be calculated. The method of Minimum Description Length is investigated for this purpose.
This paper describes the use of Disparity Space for automatic stereomatching, differential rectification and DEM production. Any matching technique which produces local (potentially ambiguous) matches can be incorporated within the Disparity Space system. A fast, optimal solution is obtained using Dynamic Programming. We have combined image intensity, local edge magnitude, local edge orientation and the output from a Laplacian of Gaussian to obtain very dense range maps. Experimental results with close range images are reported. The method has been implemented on a UNIX Workstation and is highly suitable for implementation with parallel processors.
The regularization method, which is performed by minimizing an energy functional of the image, has recently been applied to many ill-posed problems in computer vision. Notably, Grimson(1983) developed a regularization method for surface reconstruction which used a sparse set of known elevation data. We have developed an approach to surface reconstruction using both contour image data and a sparse set of known elevation values. We define a new energy functional which integrates three kinds of constraints : smoothness, fitness, and contour line constraint. These constraints seek to ensure that the reconstructed surface smoothly approximates the known elevation values and has the same height value for all points on a contour line. The energy functional can be minimized by solving a large linear system of simultaneous equations. We have successfully reconstructed a detailed 3D topography by applying this method to contour lines and known sparse elevation data extracted from moire images and topographic maps.
The physical properties of photographic roll film in conjunction with the design of small format cameras are generally optimised for ease of use and reliability during pictorial photography. Such design contributes greatly to the problems associated with the accuracy of calibration and analytical data reduction of non and semi metric small format imagery. The results of experimentation into both in-plane and out-of-plane film deformation, using a variety of commercially available 120, 220 and 70mm monochrome film stocks, are presented. Physical parameters investigated include environmental conditions, image density, processing regime, measurement and film back variations. Some results from the close range calibration of a modified small format camera are discussed in the light of the above experiments. A practical analytical data refinement system is envisaged, allowing potential accuracy to be balanced against the economics of the time consuming measurements of reseau images in an analytical stereoplotter.
For the attainment of the highest accuracies in close-range industrial photogrammetry it is necessary to account for the variation of radial distortion within the photographic field, especially for large-format lenses. In this paper an empirical correction model is introduced which overcomes serious shortcomings in the traditionally applied geometric correction approach. The new model has been tested with success on a number of medium- and wide-angle, large-format lenses. The results of sample tests are presented, and the effectiveness and practicability of the empirical approach are discussed.
This paper presents the new, microprocessor controlled, large format film camera Rol lei LFC metric. By means of rigorous reseau technique this camera defines the image space with high geometric accuracy and reliability. The camera is designed to meet the de- imands of close-range industrial photogrammetry.
This paper introduces the Linhof METRIKA as a new 4" x 5" roll film camera for close-range photogram- metry. It was aimed to develop a robust, precise and moderately priced camera which is rather handy and light- weight despite of the 4" x 5" photo format. Lens cones of 90 mm (wide angle) and 150 mm are available equipped with focussing stops which reproducibly fix a set of interior orientation parameters. Film flattening is provided by both a vacuum system and reseau technique. The design of the camera allows the interchange of reseau plates with different reseau arrangements. All operating functions of the camera are electronically controlled. First results of test field calibrations are presented which certify the high accuracy and physical stability of the camera. In addition, examples of applications of the METRIKA to architectural and industrial photogrammetry are given.
In this paper, we propose a new method for extracting vanishing points from real scenes, which exhibits linear computational complexity and good precision. This technique is very useful for camera calibration in photogramme- try, for the recovery of the motion of rigid objects and for the reconstruction of 3D scenes. The linear computational complexity is due to the introduction of the polar space, which permits the selection of the segments that converge on the same vanishing point, before computing the vanishing point itself. Extensive experimentation on real images shows that vanishing points can be identified and located even in cluttered images.
This paper considers the segmentation of range image measurements into surface patches which are either plane or curved and which are described formally by a function. After a formal description of the segmentation, we present and compare three methods suited for plane and curved patch segmentation and show the results of experiments conducted for testing their practical behaviour. The two first methods use the classical approach of region growing whereas the third method is based on a relaxation process. This original and last method exhibits simplicity and low computational complexity. Thanks to its parallel nature, it can be considered as a good candidate for range image segmentation in real-time applications.
The automatic processing of stereo images for 3D information extraction can be simplified by colour classification of the stereo image pairs prior to the correlation matching procedures. The number of corresponding pixels to be matched is drastically reduced, if only those pixels on the epipolar line are correlated which belong to the same class of colours. Colour classes of natural scenes like aerial photographies are usually composed of complicatedly shaped clusters which makes the manual setting of RGB classifier parameters an almost impossible task. We present a real-time IHS-colour classification system which uses neural network principles based on self-organizing look-up-tables for learning typical colour classes . A learning rule for the supervised training of this LUT classifier is presented . The proposed LUT classifier shows all the positive features of a 3-layer perceptron Neural Network, but performs 70.000 times faster then a simulated perceptron network and uses low-cost, commercially avai- lable components. The height measurement of plantlets for the automation in greenhouses is presented as a typical application of these principles.
A new texture feature, called Direction Measure(DM), is proposed in this paper, The new feature is based on the principle that in a homogeneous texture the gray values of pixels along any direction must change according to certain regularity( determined or stochastic); To determine such kind of regularity can get much information on the texture structure or statistical properties. The DM can also be easily extended to rotation-invariant features which can recognize the same texture no matter in what orientation it is pictured. The new rotation- invariant features are also very effective and are really rotation-invariant. Another advantage of the DM lies in that it can be easily computed and implemented by multiprocessors in a parallel way.
Softcopy photogrammetry's advent is a product of current advances in computer workstation technology and the low cost of image processing boards. Photogrammetric high resolution sensing remains, however, the realm of classical film-based photography. For softcopy-based digital photogrammetry to succeed creative solutions are needed to convert film to pixel arrays. We describe a novel scanning concept that permits one to convert a full frame of metric photography into an array of 32,000 x 32,000 pixels, or it permits the interactive "grabbing" of windows of pixels with sub-pixel accuracy while supporting a software-controlled zoom range with minimum pixels of 811.m and maximum pixels of 170p.m. At the heart of the new concept is the so-called "invisible reseau". Implementation of this new concept is in the form of the VX-series of scanners. The concept is discussed, and performance characteristics of the VX-Scanner are subject to analysis. The issue of "optimal" scanning parameters in both geometry and radiometry is a concern. The benefits derived from an interactive zoom-based approach to scanning are being highlighted.
A computer program package has been developed which combines photogrammetric numerical calculations with a graphical 3D-editor for the measurement of close range photographs. Data acquisition will be provided by scanning the images. Based on digital images, the editing of point and line information is done at the screen. The graphical editor is supported by means of bundle adjustment techniques. The developed program package will be explained and demonstrated by an example.
For the description of deformations and dummy motions during the crash, the Volkswagen AG uses 16mm high-speed drum cameras with frequencies of 400 pictures/sec. A system for a digital photogrammetric evaluation of the 16mm images has been developed. Several images of one epoch are completely digi- tized by a VIDEK Megaplus CCD-camera. After the measurement of signaled points in the digital images and a first orientation of the pictures using a graphic display, object coordinates are to be computed by a bundle adjustment. The interior orientation of the different cameras are deter- mined by simultaneous calibration. First experiences with the handling of the system, which in principle is useful for the photogrammetric evaluation of any digital images, and the precision of the point determination will be presented.
One of Kern's industrial measurement systems, SPACE, uses motorized theodolites equipped with CCD cameras in the telescope. This paper outlines practical mathmatical models, currently under evaluation, for relating image coordinates to angular pointings. The objective is a calibration method for video-theodolites which generates corrected pointings for targets imaged off-axis.
The calibration of comparators (or analytical plotters or digitizers) very often is achieved by means of a reseau of points on glass or film (grid plate) situated in contact with the measuring field of the comparator. From two sets of information the calibration data normally are deter- mined: from the comparator measurements of the reseau points and from their given control data. The procurement of high precision control data is costly. We, therefore, have developed the so far used procedure into one with only very few control data or even without any control data (= free selfcalibration). By choice of the size of the reseau and the density of marks the degree of calibration may be properly fixed. The new procedures have been successfully tested at the analytical plotter WILD AC 1 with results of high accuracy.
The increasing use of CAx techniques in the production processes and the continuously increasing pressure for quality assurance through regulated processes require a more extensive integration of measuring techniques in the manufacturing and data flow than in the past. For this reason, around 1985 the first considerations were developed in the quality assurance of the Volkswagen Organisation for a general data concept for use of coordinate measurement techniques (contact and noncontact) in CAI. The basis of all strategies was and still is a programming system for the generation of measuring programs, that are independent, i.e. excluding component specific, of the method of measurement (system AUDIMESS), as integral parts of the general system VWGEO.
The dimensional control of Agip's offshore structures by means of photogrammetry is foreseen during the structure's construction or at the prefabrication yard. The purpose of this check is to ascertain whether the parts have been built in correspondence to the design's dimensions. It can be achieved through the calculation of diameters, center distances, angles. The photogrammetric process determines the coordinates of a certain number of points readable on the surface of the structural component. Agip's control procedure involves, first of all, processing of these coordinates by means of software program in order to determine the mean radius and the pipe axes directions. Afterwards the coordinates are again processed, this time by means of the Cad system, where the "as built" model will be compared with the theoretical model, previously filed during the design processing. The comparison between the two models allows the detection of any discrepancies and their possible correction.
The objective of this paper is to present a theoretical model which will create a computer vision system capable of monitoring the fabrication of a facil- ity. The monitoring process would be able to identify out-of-position components and to re-create an as-built CAD model from the construction process. An as-built schedule would be a direct output and would provide for improved scheduling inter action and analysis. While computer vision .technology has not reached maturity where arbitrarily complex scenes can be processed and "understood", there h reason to believe that a vision system for monitoring the fabrication of a facility can be developed. To create such a vision system will require a number of tasks. The focus of this project is to use the information in the CAD system together with the camera model, the camera location, the orientation of the opti- cal axis and the positions of the illuminators to predict how the objects in the facility should appear. This information can then be used in "top-down" analysis of the scene. In a top-down analysis a scene model is used to guide the search for the objects in the scene. While developing sophisticated computer vision systems has, in general, been difficult, developing sophisticated top-down vision systems has not.
Dental medicine needs to observe the motion of the jaw with respect to the skull in three dimensions. This represents, therefore, a problem domain in which one has to observe, in real-time, the motion of one three- dimensional body in 3-D space (the jaw) with respect to another three-dimensional body in 3-D space (the skull) which both may move independently. This paper discusses an innovative solution to this requirement. The solution is implemented on a personal computer and is based on light-emitting diodes that are attached to the two moving 3-D objects. The innovation has been granted patent protection2. An element of the solution is the hand-held 3-D cursor whose position is also trackable as a separate three-dimensional body in 3-D space and allows the user to identify the XYZ coordinates of any point by a free-hand pointing action. Applications of this real-time 3-D measurement system are not only in dental medicine but may extend to mechanical engineering, medical gait analysis and other applications where 3-D motions need to be tracked in real time.
During the design/development of the Canadarm Remote Manipulator System (RMS) in the 1970's for the NASA Space Shuttle, the NRCC, National Aeronautical Establishment, Structures and Materials Laboratory initiated the development of an Online 30 Hz Video Real-Time Photogrammetry System (RPS) for Real-Time 3-Dimensional Control. With the initiation of the Canadian Astronaut Program and Canada's participation in the Space Station, a protoflight Space Vision System, based on the RPS technology, was designed and built for integration with the Shuttle Closed-Circuit Television System (CCTV). The SVS accepts one or two CCTV selected camera inputs and provides, as output, a real-time display of the photogrammetrically derived 6-degree of freedom guidance information to the RMS operator in symbolic and alphanumeric form. The SVS makes it possible, for the first time, to carry out a series of Shuttle-Manipulation tasks in the weightless/thermal/lighting environment of space using real-time guidance information. This paper describes the basic multi- processor architecture of the SVS, the photogrammetric algorithm(s) defined multiple target array payload and targeted berthing hardware, the pre-flight camera calibration set-up procedures, and the on-orbit reference calibration procedures for task related coordinate transformations to cater for thermally changing shuttle-RMS deformations. The series of experiments and graphic displays, supported by the task configuration database software of the SVS, will be summarized.
Using stereo vision and marks pasted on wall to solve the guidance and the environment learning problem for a mobile robot is proposed. A circular window is used to detect corner points as features in the scene. From the perspective transformation matrices of the two cameras and matching point-pairs in images the 3-D coordinates can be determined. With ground projections of these space points and the constraint of minimum diameter of the mobile robot, a few separated polygons are used to represent the occupied space of the surrounding obstacles. With the corner points of the mark the location of the mobile robot can be determined not only by the stereo vision system but also by each of the two cameras with plane-motion constraint. Experimental results show that the depth measurement error is less than 2%, the location determination of the mobile robot working on the ground plane is less than 5%, and the map of the occupied space can be used in further path-planning or collision avoidance.
For many tasks in turbulent flow research the simultaneous determination of tracer particle coordinates with high spatial and temporal density is required as a basis for particle tracking velocimetry (PTV). As a simultaneous, non-contact and automation-friendly method digital photogrammetry is a suitable tool to provide this information on-line. This paper describes the implementation of a system based on three CCD-cameras, which can determine the coordinates of up to 900 particles per image triplet. After an overview about the hardware used for acquisition and storage of three synchronous sequences with some hundred images each, the chain of algorithms for a completely automatic processing of the sequences is shown. These algorithms deal with some special problems arising from the task, for example the multimedia geometry, the high particle density which leads to overlapping particles in the images, and the fact that there is no continuous surface in object space, which causes ambiguities in matching. It will be shown, that these ambiguities can only be solved reliably by a system with at least three cameras.
At the Federal Institute of Technology at Zurich, an interdisciplinary research team in collaboration with the industry have developed a robot-theodolite. Such future instruments have not to be operated. They detect the target points automatically by themselves, calculate their coordinates in whatever geodetic system and telecommunicate the results to whatever spot in the field and/or a dataprocessing centre. This paper deals with the techniques of such instruments, with possible applications and with metamorphic influences on the surveyors' profession.
A submarine consists of cylindrical, spherical and conical sections, which are joined together in the construction phase. It has to be checked by measurements whether the sections have been constructed with the required precision. In order to proof to a ship- building company in the Netherlands that traditional measuring methods for quality control can be replaced by photogrammetry a test measurement has been carried out. For this test a conical submarine section has been measured twice. The first time in the traditional way and the second time by using photogrammetry. In the photogrammetric procedure retrotargets, an analytical plotter and a bundle adjustment program have been used. In this paper different measuring methods for quality control will be described shortly. The advantages and disadvantages of each method will be discussed. It will be shown in which way the required object coordinate precision of 1 mm or smaller could be obtained by photogrammetry. It is shown that curvature deviations determined with photogrammetry are almost similar with the deviations as obtained by using the traditional measuring method of the ship-building company.
The W. M. Keck Observatory will house the largest optical telescope ever constructed. A multi-segmented, 10 m diameter mirror will allow the telescope to look into space over a distance four times further than any other current system. The support cell for the mirror segments was manufactured in Spain, partially dismantled for shipment and then re-assembled and installed on the mountain of Mauna Kea in Hawaii. Photogramme.tric surveys of the structure were made both in Spain and Hawaii. Measurements in Spain served as acceptance criteria for the structure in addition to detailing as-built conditions. Loaded deformation was also tested in Spain to confirm conformance to design. The measurements in Hawaii checked for effects of shipment and also proper installation in a very different physical environment. All surveys yielded similar precision of 0.03mm RMS one sigma (1 part in 300,000 of the object diameter). This paper describes the major phases of the project, details specific design and operational characteristics of the surveys and presents the results obtained.
Stereoscopic observations of 70 mm colour transparencies have been used in bundle adjustments to determine deformations of structural walls subjected to mine subsid- ence and the Newcastle, Australia, earthquake of magnitude 5.5 on the 28th December 1989. The photography was taken with a pair of Hasselblad 500 ELX cameras which had been modified by the addition of a glass reseau plate. The cameras were fixed 2 m apart on a bar, with camera-to-object distances varying from 10 m for the mine subsi- dence tasks to 30 m for some of the earthquake analyses. The stereoscopic observ- ations were made on an Adam Technology MPS-2 Analytical Stereoplotter with only the plate co-ordinates used in the bundle adjustment. Closures of under 4 micrometres on the frames were obtained from the bundle adjustments with an extremely low rejection rate of observations being a notable feature. The unusual combination of stereo- scopic observations, an analytical stereoplotter, non-metric cameras, bundle adjust- ments and a CAD program to provide timely results, is discussed.
The optimum determination of the position of projective centres in industrial photogrammetry can be solved twofold: On the one hand simulations can contribute considerably to solve the problem, and on the other hand analytic formulations might be used. For reasons of powerful network design the latter ones are preferred because here different objective functions can be integrated in one solution strategy only. The paper reviews the state-of-the-art in analytic first order design. It introduces criterion matrices for the coordinates, which are the 'observations' for solving the design matrices. The algorithm used during optimization is the Lemke algorithm. That means, the optimization is performed by quadratic programming being in correspondence with linear complementarity problems. Besides the objective functions for coordinates further boundary conditions have to be considered such as constraints in accuracy, position and costs.
Recent developments of video cameras and frame grabber technology have led photogrammetrists into the fields of digital image processing and machine vision. This paper describes the components of a digital photogrammetric system for close range applications. The approach to image matching involves the location of preliminary matched details based on features which are used as initial approximations to determination of precisely matched areas on the two images. Image matching based on grey scale values uses the least squares matching method. A software package has been developed to determine object geometry from overlapping digital images, and results from test are presented for three combinations of base/height ratio. Future developments will be based on the use of a knowledge base to control the searching for matched points.
A system is presented that allows maps compilation in a fully digital way: digital image and D.T.M. are used in order to obtain an orthoimage; it will be the base for numerical and graphical map compilation. System concepts are discussed and some applications presented.
There is increasing use of electronic sensors and digital signal processing for measurements of optically acquired data. Applications include automatic inspection, surveying, remote sensing and photogrammetry. Sensors, at the present time, require subpixel methods to improve the resolution above that available given the spacing of sensing elements and the analogue to digital conversion resolution. This paper reviews proposed subpixel methods in the context of an increasingly important application, namely, the determination of the position of a laser spot on a sensing array for triangulation. A number of techniques are chosen and analysed experimentally. Their performances are compared and contrasted with respect to spatial resolution, quantisation accuracy and noise. For the comparison, use is made of simulated data, and real data obtained from a triangulation system9.
A new approach to reconstruct the three-dimensional surface of the object space from digital images is described. All the object points obtained by an automatic orientation procedure lead to a first approximation of the surface. Edges are computed for one image and matched to the other image by grey level correlation or least-squares matching through the scale space. To every discrete step in the scale space there exists the digital stereopair (image pyramid), the corresponding surface (digital elevation model DEM) and the warped images. The warped images in this discrete scale space representation correspond to digital orthophotos obtained from the DEMs that result from matching the image pyramid. We propose to use the warped images on every successive level in the image pyramid in order to reduce the foreshortening problems associated with any area-based matching method. The paper describes the method and some experimental results are reported.