At The Center for Mapping of The Ohio State University, a mobile mapping system was developed. By integrating the global positioning system (GPS) with a stereo vision system, the global coordinates of any point in the field of view of the cameras can be obtained. The calibration of cameras was done by using the bundle adjustment in a test field of retro reflective targets. The image coordinates are measured automatically by means of the center of gravity, the identification of points is also performed automatically using the space resection. As the stereo cameras are mounted on a stationary platform, relative orientation constraints are applied in the calibration. As a result, the number of required observations is reduced and a stable calibration of the stereo vision system is achieved. The modifications to the standard bundle solution are presented, as well as the positioning accuracy of the calibrated system.

This paper presents a new simple practical method for estimating the horizontal scale factor and/or the pixel aspect ratio of a standard imaging system that uses a CCD or CID video camera. The method uses a self-aligning target pattern that can be produced by a conventional 300 dpi laser printer. It requires only three one-dimensional measurements from a single image - one of which is the null of an interference pattern - and does not depend on exact knowledge of the focal length of the lens or the working distance of the camera. It does not require the use of an oscilloscope or removal of the anti-aliasing filter on the output of the camera. In addition to finding the scale factor, the same target pattern is useful for orienting the camera about its optical axis.

This paper presents a new approach for calculation of the extrinsic parameters of a camera using coplanar calibration points. The basic principle of the method is as follows: first create the mapping (MP) from the plane on which the calibration points lie to the image plane, and the extrinsic parameters are then calculated on the basis of MP and vanishing point concept. Experimental results showed that the method is effective.

An image-based technique for passive acquisition of 3-D range data is proposed and tested. The distance is extracted, in this technique, from the estimation of focus conditions on images produced through a monocular imaging system under natural illumination. To estimate the depth, 3 - 4 images need to be taken under camera settings with known associated parameters. The 3-D surface geometry is acquired by applying the algorithm in parallel to all surface points in the field of view.

The objective of this work is to develop a technique for 3D shape information recovery from multiple 2D images. Stereopsis is one of the most popular techniques for recovering 3D information. But it has an inherent and difficult problem with stereo matching, especially when the epipolar line is not parallel to the raster line. In this paper, we propose a new technique to avoid this problem by projecting back the feature points in images to the 3D space and by voting (just like for the Hough transform) with a well-contrived voting and evaluation rule. The octree representation of the object is adopted to enable multi resolutional analysis and to increase calculation efficiency. Experimental results are also described.

A mobile mapping system which integrates the global positioning system (GPS) and stereo cameras was developed by the Center for Mapping of the Ohio State University. This system was calibrated using the bundle adjustment with the relative orientation constraints. To extract the useful data from those images, a three-step image matching method based on the epipolar geometry was developed: first, the approximated position is estimated from the system geometry; second, the cross correlation method with the significant maximal coefficient was used. A variable size of template is employed to find the significant maximal correlation coefficient. Third, least squares matching was used to have the sub pixel accuracy. Using this matching method, the point, profile and surface can be measured by selecting points in a single image of a stereo pair. Finally, this technique is extended to extract three dimensional edges by line following. All data captured by the mobile mapping system are available in a global coordinate system.

Moire projection method enables non-contact measurement of the shape or deformation of different surfaces and constructions by fringe pattern analysis. The fringe map acquisition of the whole surface of the object under test is one of the main advantages compared with 'point by point' methods. The computer analyzes the shape of the whole surface and next user can selected different points or cross section of the object map. In this paper a few typical examples of an application of the moire technique in solving different medical problems will be presented. We will also present to you the equipment the moire pattern analysis is done in real time using the phase stepping method with CCD camera.

Detection of object points on non-textured surface are one of the basic image processes in some photogrammetric application. This paper deals with the testing method and its relevant theory on the determination of the coordinates the object points in photogrammetric test field. It includes real-time positioning and surveying system with a CCD camera, data preprocessing, photogrammetric processing and the test of determining the coordinates object points of the test field in real time.

In industrial measurement applications a great variety of tasks require the measurement of non-targeted ('natural') objects. Edges are important features of industrial objects. Visualization and texturing of objects is often achieved by structured light. This paper investigates an alternative approach based on diffuse illumination. It describes an extension of the Multi-Photo Geometrically Constrained (MPGC) matching algorithm which allows the dimensional positioning of these features. The algorithm is applied to edge tracking. The major algorithmic aspects will be addressed. Practical data from a demonstration object will be used in order to show potential and problems of this approach.

A personal computer based vision metrology system for non-contact measurement of nominally two-dimensional parts has been developed. Through use of the reflected image resulting from light incident at grazing incidence, it is possible to determine the correct dimensions even if burrs or other manufacturing defects are present on the end of the part. Thus the system can, for example, define the complete profile of short samples cut from a long aluminum extrusion without special sample preparation before measurement. In this paper the application of the technique to circular tubes is described in some detail and factors involved in its extension to more complex shapes are discussed.

A method of displaying time dependent visual phenomena in a single image is discussed. Use of hybrid video and digital processing provides fast and economical visual data accumulation over short or long periods of time.

Readily available solid-state cameras range from inexpensive camcorders to specialized systems with cooled 'high-resolution' sensors. Prices of such systems are between $DOL500.- to $DOL100,000.-. The choice of an appropriate imaging system, both in price and performance, is thus of interest to system developers and users. Although the radiometric and geometric characteristics of these systems vary drastically, the degree to which these differences affect the accuracy of three-dimensional measurements has not been investigated in detail. This paper compares the three-dimensional accuracy of some representative solid-state cameras (S-VHS camcorder, off-the-shelf CCD-cameras, 'high- resolution' camera) via a testfield with highly accurate reference coordinates. The performance of the different systems is analyzed with respect to potential limiting factors and improvements which could enhance the accuracy. Accuracies exceeding 1/20^th, 1/85^th, and 1/15^th of the pixel spacing were attained with a S-VHS camcorder, an off-the-shelf CCD-camera, and a 'high-resolution' camera with 3072 X 2320 pixels. The high level of performance is also evidenced by the relative accuracies in object space of 1:15000, 1:50000, and 1:35000 for the respective cameras.

Many industrial metrology applications require accurate and real or near-real time measurement tools in often adverse conditions. Over the past three years, the Industrial Alignment Project (IAP) has been focused on the development of precise measurement techniques for use with large rotating machinery. This paper highlights some results using digital array cameras which have been shown to be well suited for many applications. Limited camera resolution and high accuracy point positioning requirements have required the development of specialized targets. Retro-reflective targets are used to reduce the varying effects of lighting/shadows since this material exhibits maximum reflection in the direction of the light source. Using sub-pixel edge detection techniques with the pre-defined targets, it is possible to obtain accuracies of 1/20th a pixel or better. In addition, employing multiple camera stations with an appropriate imaging geometry, the three dimensional coordinates of the point of interest can be obtained by establishing photogrammetric techniques. The results of test projects are given to illustrate the viability of such a measuring system.

Automated photogrammetric vision systems, called photogrammetric stations, are used for industrial on-line control. The stations are on-site calibrated camera setups with necessary image processing in order to provide the manufacturing process with three-dimensional control data. One of the first operative industrial applications is the car body orientation within a seam sealing cell in automotive manufacturing.

The Offshore Oil and Gas Industry in the North Sea has many requirements for three- dimensional measurements in air and underwater. A market audit found that use of conventional photogrammetry was being rejected for many applications because the information was not available fast enough. A development project was set up to replace the photographic cameras with a choice of video or high resolution digital electronic cameras, and the analysis system with a personal computer based image processing system. This solution is now in operation. The paper details the in-house development of the high resolution digital electronic camera and the personal computer based measurement hardware and software. It includes a discussion of the technological parameters, including the method for pixel for pixel correlation within the digital system, camera calibration techniques, the system algorithms, sub-pixel measurement and dimensional accuracy. It introduces the work that was carried out to make the final product acceptable to structural engineers, who now use it to transfer three- dimensional measurements to their CAD systems. It also looks at the work that is being carried out to transform the system into a closed loop control system for underwater robotic manipulators, which includes binary conversion, convolution filtering and tracking functions.

The most relevant parameters to control the quality of glass vials are the internal and external diameters of the mouth and the height of the rim. A low cost vision system based on a 486 PC, a frame grabber, 4 CCD cameras (768 X 512 pixels) and I/O device to control the production of vials, by adjusting the flames temperature in the moulding section of the machine, has been developed and tested. A 24 mandrel machine rotating at about 300 rpm with a production capability of about 4200 pieces/hour had to be monitored with an accuracy of +/- 0.02 mm in the measure of the mouth diameters and +/- 0.04 mm on the rim height. In order to minimize the time delay required for the machine temperature compensation, the measurements had to be taken during the forming process. The system must be fast enough to follow the process, able to take into account the temperature variation of different classes of vials and far enough from the high temperature of the flames. A direct calibration procedure, using a reference vial, and a pyrometer to check the temperature range was derived. A long focus lens coupled with a bellow to put the system away from the flames was adopted. The algorithm implemented for the measurements and the machine temperature control is presented.

Recent progress in establishing calibration schemes for video cameras promises to significantly enhance the measurement capabilities available to aeronautical engineers studying flight dynamics. Targeting and lighting conditions in working facilities are usually less than ideal from a photogrammetric standpoint. This paper will discuss the implementation of a customized state-of-the-art commercial tracker used to estimate pitch attitude and spin rate of 10 Hz refresh rates for a free spinning model in a 20 foot diameter vertical wind tunnel. Active efforts to augment the measurement accuracy of the real time system by post processing concurrent, time coded, taped data from two independent camera stations will also be described. Independent calibration of interlaced fields, degradations due to storage media, effects of electronic shuttering, limited numbers of target points, estimation of 'optical flow' of data from field to field, vibration, and conflicting requirements for the competing video systems will be discussed.

Applications such as dimensional inspection, positioning and tracking of objects, and reverse engineering require highly accurate measurement systems. A vision-based coordinate measurement system (VCM) developed at the National Research Council of Canada, targeted for these applications, will be described. In order for a vision system to be considered for such applications, its performance and design parameters must be well understood. Therefore, a performance evaluation procedure and accuracy analysis, along with test results, will also be presented.

Stereo vision system based on the principle of triangulation has been built for shape inspection of technical object. It consists of two CCD-cameras, a He-Ne laser, a special electronic circuit and a personal computer. Here, we propose a method of non-coplane of every three calibration points and the optical center of camera to calibrate the system. By making some measures, we can improve the measurement accuracy greatly.

Coherent optical communication offers very high sensitivity and enormous bandwidth as compared with the existing noncoherent system. Unfortunately it is still under experimentation and research. With the anticipation of voluminous communication, it is predicted that coherent systems will play a significant role in future applications. This paper will develop a model for a coherent frequency modulated video receiver using heterodyne detection and will highlight the potential signal to noise ratio improvements over the noncoherent receiver. The requirements to achieve such a high quality system will also be emphasized.