The adaptation of an image dissector tube (IDT) within the OPTFOLLOW system provides high resolution displacement measurement of a light discontinuity. Due to the high speed response of the IDT and the advanced servo loop circuitry, the system is capable of real time analysis of the object under test. The image of the discontinuity may be contoured by direct or reflected light and ranges spectrally within the field of visible light. The image is monitored to 500 kHz through a lens configuration which transposes the optical image upon the photocathode of the IDT. The photoelectric effect accelerates the resultant electrons through a photomultiplier and an enhanced current is emitted from the anode. A servo loop controls the electron beam, continually centering it within the IDT using magnetic focusing of deflection coils. The output analog voltage from the servo amplifier is thereby proportional to the displacement of the target. The system is controlled by a microprocessor with a 32kbyte memory and provides a digital display as well as instructional readout on a color monitor allowing for offset image tracking and automatic system calibration.
The Ariel Performance Analysis System is a computer-based system for the measurement, analysis and presentation of human performance. The system is based on a proprietary technique for processing multiple high-speed film and video recordings of a subject's performance. It is noninvasive, and does not require wires, sensors, markers or reflectors. In addition, it is portable and does not require modification of the performing environment. The scale and accuracy of measurement can be set to whatever levels are required by the activity being performed.
The Videomex-X is a new product intended for use in biomechanical measurement. It tracks up to six points at 60 frames per second using colored markers placed on the subject. The system can be used for applications such as gait analysis, studying facial movements, or tracking the pattern of movements of individuals in a group. The Videomex-X is comprised of a high speed color image analyzer, an RBG color video camera, an IBM AT compatible computer and motion analysis software. The markers are made from brightly colored plastic disks and each marker is a different color. Since the markers are unique, the problem of misidentification of markers does not occur. The Videomex-X performs realtime analysis so that the researcher can get immediate feedback on the subject's performance. High speed operation is possible because the system uses distributed processing. The image analyzer is a hardwired parallel image processor which identifies the markers within the video picture and computes their x-y locations. The image analyzer sends the x-y coordinates to the AT computer which performs additional analysis and presents the result. The x-y coordinate data acquired during the experiment may be streamed to the computer's hard disk. This allows the data to be re-analyzed repeatedly using different analysis criteria. The original Videomex-X tracked in two dimensions. However, a 3-D system has recently been completed. The algorithm used by the system to derive performance results from the x-y coordinates is contained in a separate ASCII file. These files can be modified by the operator to produce the required type of data reduction.
Motion analysis studies require the precise tracking of reference objects in sequential scenes. In a typical situation, events of interest are captured at high frame rates using special cameras, and selected objects or targets are tracked on a frame by frame basis to provide necessary data for motion reconstruction. Tracking is usually done using manual methods which are slow and prone to error. A computer based image analysis system has been developed that performs tracking automatically. The objective of this work was to eliminate the bottleneck due to manual methods in high volume tracking applications such as the analysis of crash test films for the automotive industry. The system has proven to be successful in tracking standard fiducial targets and other objects in crash test scenes. Over 95 percent of target positions which could be located using manual methods can be tracked by the system, with a significant improvement in throughput over manual methods. Future work will focus on the tracking of clusters of targets and on tracking deformable objects such as airbags.
Analysis of the body in motion is often necessary to understand the functionality of the subject. In real-time observation, the subtleties between proper physical functions and deviations from the norm or desired state may not be seen due to the short time duration that the motion occurs within. Expanding time by the use of high-speed video imaging can improve observational data gathering methods and data accuracy.
The recent introduction of a two dimensional interactive software package provides a new technique for quantitative analysis. Integrated with its corresponding peripherals, the same software offers either film or video data reduction. Digitized data points measured from the images are stored in the computer. With this data, a variety of information can be displayed, printed or plotted in a graphical form. The resultant graphs could determine such factors as: displacement, force, velocity, momentum, angular acceleration, center of gravity, energy, length, angle and time to name a few. Simple, efficient and precise analysis can now be quantified and documented. This paper will describe the detailed capabilities of the software along with a variety of applications where it might be used.
While much of the software developed in research laboratories is narrow in focus and suited for a specific experiment, some of it is broad enough and of high enough quality to be useful to others in solving similar problems. Several biomechanical assessment packages are now beginning to emerge, including: * 3D research biomechanics (5- and 6-DOF) with kinematics, kinetics, 32-channel analog data subsystem, and project management. * 3D full-body gait analysis with kinematics, kinetics, EMG charts, and force plate charts. * 2D dynamic rear-foot assessment. * 2D occupational biomechanics lifting task and personnel assessments. * 2D dynamic gait analysis. * Multiple 2D dynamic spine assessments. * 2D sport and biomechanics assessments with kinematics and kinetics. * 2D and 3D equine gait assessments.
Northern Digital's first active marker point measurement system, the WATSMART, was begun in 1983. Development ended in 1985 with the manufacture of a highly accurate system, which achieved .15 to .25 mm accuracies in three dimensions within a .75-meter cube. Further improvements in accuracy were rendered meaningless, and a great obstacle to usability was presented by a surplus light problem somewhat incorrectly known as "the reflection problem". In 1985, development of a new system to overcome "the reflection problem" was begun. The advantages and disadvantages involved in the use of active versus passive markers were considered. The implications of using a CCD device as the imaging element in a precision measurement device were analyzed, as were device characteristics such as dynamic range, peak readout noise and charge transfer efficiency. A new type of lens was also designed The end result, in 1988, was the first OPTOTRAK system. This system produces three-dimensional data in real-time and is not at all affected by reflections. Accuracies of 30 microns have been achieved in a 1-meter volume. Each two-dimensional camera actually has two separate, one-dimensional, CCD elements and two separate anamorphic lenses. It can locate a point from 1-8 meters away with a resolution of 1 part in 64,000 and an accuracy of 1 part in 20,000 over the field of view.
Experience with a turnkey analysis system featuring high resolution video input and display, a modular video disc
system and a 16 mm cine film scanner with 2600-point resolution, is presented. Tracking is performed with a high-speed
correlation process, requiring no special markers. Software packages for evaluating two and three-dimensional results are
interactively accessible. Combining the original image sequence with real-time graphic overlays and active drawing of
graphic diagrams, provides for an excellent understanding and documentation of the motion sequences.
Biomechanical analysis of motion is based on the approximation of skeletal segments as rigid links moving through space, interconnected through a series of low-friction joints. Measurement systems that are aimed at capturing the spatial trajectories of body segments usually involve a camera system that tracks a series of body-fixed markers. Using stereophotogramxnetric cameras, the planar projections of markers at each camera are used to reconstruct the spatial coordinates of each marker. The derivation of segmental kinematics (i.e. linear translation and angular orientation) necessary to document the motion of the body segments has been done in most cases by attaching the markers to anatomic landmarks, and using geometric assumptions to characterize the spatial motion of given limbs. For example, by tagging the hip, knee and ankle statements have been made about the motion of the knee joint, and therefore the shank and thigh segments. This approach suffers from serious shortcomings, including the approximation of body segments as lines (as opposed to rigid bodies), the underlying assumption that axes of rotation remain constant throughout the motion, and that joint centers can be tagged by skin-mounted markers.
The motions studied In biomechanics today are Increasingly more complex, requiring spatial, force and myographic measures. The Peak system employs off-the-shelf video equipment and an AT compatible microcomputer to perform spatial and analog data acquisition. Every video field can be analyzed in either NTSC (North American) or PAL (European) video formats. Video sampling rates can vary from 50 to 2000Hz, depending upon the video hardware used. The body locations of interest may be obtained using any of three methods. The manual acquisition method superimposes a colored cursor over the video image and its position is controlled by a handheld mouse. The semi-automatic option controls the initial position of the cursor by predicting the position based upon its location in previous frames. The value of these options is that they can be applied in difficult experimental settings or for digitizing points that are internal to the body (e.g., joint centers). The third method is automatic tracking, which determines the center of area of a marker that contrasts with its immediate background. An individual criterion may be set for each marker to discriminate it from noise, background and other markers. Hidden markers can be treated via manual digitization, interpolation or ignored if seen by other cameras. Once collected, spatial data can be translated and/or rotated with respect to external reference frames, or to internal reference frames within the moving subject. An analog to digital sampling module acquires data from sensors and synchronizes them with the video generated spatial data. Subsequently, complete motion information can be generated rapidly and accurately for countless applications. The performance of the video measurement system was assessed by measuring its precision and accuracy in two dimensional static tests. The precision for manual digitizing was found to be 1 in 2422 and for automatic digitizing, 1 in 5280. The accuracy for manual digitizing was found to be 1 in 3267 and for automatic digitizing, I in 4310.
The SELSPOT 3D Motion Measurement System provides accurate measurement data f many kinds of movement. The SELSPOT II camera and MULTILab software package can be used to perform a complete analysis of the data. The camera has a patented photodetector which registers light pulses emitted by LEDS attached to the object being analyzed. Because the LEDs (markers) are sequenced by the system, there is no "guessing" as to which marker went where or which markers were obscured. The electronic system automatically adjusts the light intensity of the LEDs and suppresses ambient light. The camera registers the X and Y coordinates of the signal, digitizes the data, and then sends the data to the computer for analysis and reporting. 3D analysis is possible with two or more cameras. Up to 16 cameras and 120 LEDs can be used. The base sampling rate is 10 kHz -- for example, with two cameras and five LEDs the sampling rate would be 1000 Hz. Resolution is 0.025% of the measuring range; non-linearity of the detector is less than 0.1% of the measurable range. The software also allows input from up to 48 analog channels supporting analog sensors, such as, force plates, EMG electrodes, etc. The data from the camera(s) and other sensors are synchronized by the system. MULTILab is a high-level programming language that performs all of the calculations for angles, velocities, acceleration, moments, etc., and provides customized reports containing all of the information necessary for sophisticated motion analysis. The system is rapid, accurate, reliable, and easy to use.
This paper describes an application-specific engineering workstation designed and developed to analyze motion of objects from video sequences. The system combines the software and hardware environment of a modem graphic-oriented workstation with the digital image acquisition, processing and display techniques. In addition to automation and Increase In throughput of data reduction tasks, the objective of the system Is to provide less invasive methods of measurement by offering the ability to track objects that are more complex than reflective markers. Grey level Image processing and spatial/temporal adaptation of the processing parameters is used for location and tracking of more complex features of objects under uncontrolled lighting and background conditions. The applications of such an automated and noninvasive measurement tool include analysis of the trajectory and attitude of rigid bodies such as human limbs, robots, aircraft in flight, etc. The system's key features are: 1) Acquisition and storage of Image sequences by digitizing and storing real-time video; 2) computer-controlled movie loop playback, freeze frame display, and digital Image enhancement; 3) multiple leading edge tracking in addition to object centroids at up to 60 fields per second from both live input video or a stored Image sequence; 4) model-based estimation and tracking of the six degrees of freedom of a rigid body: 5) field-of-view and spatial calibration: 6) Image sequence and measurement data base management; and 7) offline analysis software for trajectory plotting and statistical analysis.
Traditional 16mm and 35mm high-speed film-based data analysis is significantly improved with the introduction of a new generation, low-cost and fully integrated rear projection motion analyzer. The X-Y data are digitized and collected by an operator using a hand-held cursor. However, all film handling functions are fully automated under the control of a built-in microprocessor and/or external host computer by way of a bi-directional RS-232 port. A companion PC-based software package integrates the data collection task, minimizes operator stress and provides complete solution data. This new analyzer system provides scientists and engineers with a much faster means of digitizing and greater system accuracy when compared to existing film readers in the field.
The development of the Random Access Tracking System was initiated at the University of Muenster, Department of Orthopaedic Physiology by Dr. Theysohn. This system is a real-time high-speed and high-resolution multi-point tracking system. The moving objects are identified with retro-reflective markers which are illuminated by halogen spotlights placed around the camera lens. The video interface generates deflection signals which are fed to unique Random Access Cameras manufactured by Hamamatsu Corporation. These signals perform high speed window scanning and can sample up to 7,500 markers per second. Under certain circumstances this can be increased to 15,000 markers per second. From 1 to 126 markers can be detected in a line scan search mode. Window size may be varied in steps from 0.5% to 4.0% of the field of view. Using a small window it is possible to obtain 1 part in 32,768 in each direction of the field of view. The raw data are reduced to 2-D centroids of the targets. On-line data storage and display are possible using an industry-standard ATPC with DMA interface. Real-time feed-back is also possible. The video interface provides for off-line 3-D reconstructions using the data from two or more synchronized cameras. The system can be adapted to meet the needs of particular applications by modifying sample-rate, data transfer rate, and the number and the dimensions of the windows.
In the early 1980's, with the advent of high speed Very Large Scale Integrated Circuits, real-time image processing and pattern recognition became possible. These techniques can be used to significantly improve the performance of movement analysis equipment by increasing the reliability of marker detection. It has been shown  that the use of the cross-correlation algorithm for shape recognition improves the accuracy of marker coordinates. Furthermore, when shape recognition is used, smaller targets can be tracked without degradations in performance. The ELITE system employs a two-level, hierarchical, structure to recover 3D coordinates from images. The first level of this structure is contained in the system hardware which processes the data using a Single Instruction Multiple Data approach. This permits an unlimited number of markers to be detected in each frame, at up to 100 frames per second (Hz). The second level of the structure is contained in the system software installed in a general purpose personal computer (PC). This level corrects image distortions; establishes marker identities; performs the required tracking, space resection and intersection; filters the data stream; and provides graphic displays of results. The overall accuracy of the 3D system is estimated to be 1 part in 2800 over the field of view .
The CODA 3-D motion analysis system has evolved from initial design studies which were carried out in 1970. This paper traces the history of the technical evolution of the system from the setting of the first design criteria to its present commercial form. Attention is paid to the way in which the design was affected by advances in available technology and perceived user need over this timescale. The operating principles of the current system are described in detail and related in quantified form to the performance specifications. The contributions of various types of measurement error, both intrinsic and extrinsic to the system are analysed. Integration of 3-D movement tracking with force platform output and EMG to provide a full biomechanics measurement and analysis capability is described. Applications of the system in clinical measurement are described and typical data presented.
The PRIMAS system derives from a long line of development at Delft University of Technology , originating from  with subsequent innovations such as strobed illumination (1974) of reflective markers, to obtain the simultaneous, equidistant, periodic sampling of all marker positions; real-time estimation of the marker centroids from the full, digitized, contours (1984) to retain the on-line data reduction, while enhancing the resolution; interfacing to industry-standard AT type personal computers, with modest disk requirements and no buffering, even for long data runs; 100 Hz, 0.1 ms integration time, electronically-shuttered TV cameras, to get an optimum marker contrast in high ambient or outdoor light conditions (1986). System specifications include a precision of typ. 1:18000 (X) for 2-D coordinate noise or repeatability. With the 100 Hz sample rate this implies an unprecedented spatio-temporal resolution . This favors 3-D reconstruction, as well as a low noise propagation in the estimation of first and higher order derivatives, as are routinely required in biomechanics analysis. The latest feature is real-time marker identification by a software module within the data acquisition program. This option, for the not too complex situations, is feasible only by the data reduction inherent in on-line marker centroid processing. The 3-D calibration, reconstruction and further analytical and display programs are available in the ASYST 3.2 Scientific Language System. A source code option caters for customer extensions. The internal VME/VSB system bus allows the basic dual or quad camera 3-D systems to be readily expanded to larger configurations.
In countless experimental and R&D laboratories, throughout the world, the need for precise, non-contact motion measurement is evident. The objective of this discussion is to examine the need for non-contact measurement and to become familiar with an electro-optical method for obtaining precise, reliable displacement data -- without contact. Theory of operation, key specifications, capabilities and limitations of this technique are discussed. In conclusion, the electro-optical displacement follower offers a practical solution to many measurement problems. Its flexibility and ease of set-up have established it as an extremely useful laboratory tool.
COSTEL is an integrated, automated system for the measurement of human motion. It is based on solid state cameras equipped with linear image transducers (2048-element CCD arrays) and toroidal lenses1. Each camera must be considered a one-dimensional sensor, which estimates the spatial position of a plane passing through the target point and the nodal axis of the lens. Thus, three cameras (3D unit) are necessary and sufficient to determine the spatial coordinates of a target Each 3D unit is connected to a personal computer (PC) by an interface board. The target points whose positions are to be measured (markers) consist of infrared light-emitting diodes (IR-LEDs) which are fired one at a time, in a sequence synchronized with the system operation. The calibration of a 3D unit is carried out by means of a frame, formed by 0.8 m aluminum bars, supporting up to 20 markers whose positions are known with precision. The use of this small (0.8 x 0.7 x 0.7 m) and light (5 Kg) calibration object is permitted by the previous, factory made, assessment of the optical distortion error associated with each camera. The various system operations (calibration, data collection, 3D marker trajectory reconstruction, etc.) are driven by suitable PC software. Application programs are available, permitting the biomechanical analysis of lower limb joints. The COSTEL system is able to track up to 20 markers at 125 Hz sampling frequency, or 10 markers at 250 Hz. The spatial resolution is 0.025% of the measurement field-width, the precision is 0.02%, the accuracy (maximum absolute error) is 0.1%. In an upgraded version of the system the resolution will be better than 0.01%, precision and accuracy will be enhanced accordingly. An outstanding feature of COSTEL is its low sensitivity to spurious light, reflections and other environmental noise. This feature is due to the reflection coefficients at the marker radiation light wavelength (820 rim), and to the camera signal conditioning.
Most TV-based motion analysis systems, including the original version of 1/ICON, produce 3D coordinates by combining pre-tracked 2D trajectories from each camera. The latest version of the system, VICON-VX, uses totally automatic 3D trajectory calculation using the Geometric Self Identification (GSI) technique. This is achieved by matching unsorted 2D image coordinates from all cameras, looking for intersecting marker 'rays', and matching intersections into 3D trajectories. Effective GSI, with low false-positive intersection rates is only possible with highly accurate 2D data, produced by stable, high-resolution coordinate generators, and incorporating appropriate compensation for lens distortions. Data capture software and hardware have been completely redesigned to achieve this accuracy, together with higher throughput rates and better resistance to errors. In addition, a new ADC facility has been incorporated to allow very high speed analog data acquisition, synchronised with video measurements.