[Back to Index] [Back to Personal] [Back to Homepage] |
|
| Application of High Technology to Performance AnalysisGideon 'B. Ariel ' Ariel Performance Analysis System, Inc. 6 Alicante, Trabuco Canyon. California 92679 ABSTRACT Biomechanics is the study of the motion of living things and, as an established disciplines it has evolved from a fusion of the classic disciplines of anatomy, physiology, physics, and engineering. New technological enhancements now available may provide the tools to accurately describe and quantify motion exercises work environments, and/or any motion related protocols. MOTION ANALYSIS The Ariel Performance Analysis System is a computer-based system to measures analyzes and present movement characteristics. This Performance Analysis System provides a means to quantity motion utilizing input information from any or all of the following mediums: visual (video or film), electromyography (EMS), and force platforms. The theoretical bases of the system models the human body as a mechanical system of moving segments upon which muscular! gravitational! inertial, and reaction forces are applied. Although the physical and mathematical model for such a system is complex! it is well defined (isbs7). The Ariel Analysis System provides a means of measuring human motion based on a proprietary technique for the processing of multiple high-speed film or video recordings of a subject's performance (2!12!13). This technique demonstrates significant advantages over other common approaches to the measurement of human performance. First, except in those specific applications requiring EMS or kinetic (force platform) data, it is noninvasive. No wires! sensors, or markers need be attached to the subject. In fact! the subject need not be aware that data is being collected. Seconds it is portable and does not require modification of the performing environment. Cameras can be taken to the location of the activity and positioned in any convenient manner so as not to interfere with the subject. Activities in the ,workplaces homes hospital, therapist's offices health club, or athletic field can be studied with equal ease. Thirds the scale and accuracy of measurement can be set to whatever levels are required for the activity being performed. Camera placements lens selections shutter and film speed may be varied within wide limits to collect data on motion of only a few centimeters or of many meters! with a duration from a few milliseconds to a number of seconds. Video equipment technology currently available is sufficiently adequate for most applications requiring accurate motion analysis, although special applications may require very high-speed cameras, powerful lenses, and high levels of illumination. Determination of the problem, error level, degree of quantification, and price will all affect the input device selection. Fourth, in studies with subjects who are aware of the filming process, multiple trials with each individual can be collected with ease. Multiple trials and subjects enhance the development of biomechanical models, task descriptions, and performance protocols as well as allowing statistical analyses. A typical kinematic analysis consists of four distinct phases -- data collection (filming), digitizing, computation, and presentation of the results. Data collection is the only phase that is not computerized. "In this phase, film or video recordings of an activity,are made using two or more cameras with only a few restrictions: (1) All cameras must record the action simultaneously; (2) the cameras must not move during the activity or between the recording of the activity and the recording of the calibration points: (3) the activity must be clearly seen throughout its duration from at least two camera views! (4) the location of at least six fixed noncoplanar points visible from each camera view (calibration points) must be known. These points need not be present during the activity as long as they can be seen before or after the activity. Usually they are provided by some object or "apparatus" of known dimensions that is placed in the general area of the activity, filmed and then removed! (5) the speed of each of the cameras (frames/second) must be accurately known, although the speeds do not have to be the same; and (6) some event or time signal must be recorded simultaneously by all cameras during the activity in order to provide synchronization. These rules for data collection allow great flexibility in the recording of an activity. Information about the camera location and orientation, the distance from camera to subject, and the focal length of the lens is not needed. The image space is "self-calibrating" through the use of calibration points that do not need to be present during the actual performance of the activity. Different types of cameras and different film speeds can be used and the cameras do not need to be mechanically or electronically synchronized. The best results are obtained when camera viewing axes are orthogonal (9U degrees apart), but variations of 20 to 30 degrees can be accommodated while introducing almost negligible error. Digitizing is the second phase of analysis. Initially, the video image is captured by the computer and stored in memory. This eliminates any further need for the video apparatus. The image sequence is then retrieved from computer memory and is displayed, one frame at a time, on the digitizing monitor. The grabbed image can be enhanced or altered in several ways. These include zooming the whole frame or a defined, isolated portion of the view. Changing the size may help the person digitizing to more accurately.determine a particular joint which in the unenlarged viewcould not be identified. It is also possible to adjust the coloring, shading, and intensities of the image for each frame. Thus, alterations of the video image, in a manner similar to camera adjustments, can assist in the digitizing process. Following the selection and storage of the video image, the digitizing phase continues. Using a video cursor, the location of each of the subject's body joints (e.g. ankle, knee, hip, shoulder, elbow) is selected and entered into the computer. As each point is selected, it is displayed within the video image on the monitor. The person digitizing may select an additional option which causes the previous one or two digitized frame points to appear while the current frame is being traced. This allows the digitizer .to evaluate. his current joint selection choice with the perspective of the previously identified joint center. Another useful feature is the "rubber band" effect. This situation occurs when the previously selected joint in the current frame is "fixed" while the next joint is being located. The computer connects the preceeding joint center with the curser. As the curser is moved to establish the joint center location, a line is created which moves according to the curser motion. This assists in digitizing since the connection between the two points represents the segment between the two joints. The digitizer can examine the alignment of the segment based on the perceived joint center while making the current joint center choice. Another option available when there is difficulty locating a joint center is the "missing" and "estimated" point notation. If the joint center is completely blocked from view and no appropriate estimate can be made, the particular point can be designated as "missing" and the subsequent processing phase will ignore it. If the digitizer believes that the location is approximately correct, the joint center can be designated as "estimated" with subsequent processing utilizing a weighting factor in calculating that point. In addition, a fixed point, which is a point in the field of view that does not move, is digitized for each frame as an absolute reference. This allows for the simple correction of any registration or vibration errors introduced during recording or playback. At some point during the digitizing of each view, a synchronizing event must be identified and, additionally, the location of the calibration points as seen from that camera must be digitized and saved in the computer memory. This sequence of events is repeated for each camera view. Digitizing is primarily a manual process. It is performed, however, under computer control and the digitizing of video images is computer assisted. An automatic digitizing option is also available. Automatic digitizing can be performed utilizing brightness, color resolution, and kinetic parameters. However, under manual control, user participation in the digitizing process, however, provides an opportunity for error checking and visual feedback which rarely slows the digitizing process adversely. A trained operator with a reasonable knowledge of anatomy and a consistent pattern of digitizing can rapidly produce highquality digitized images. Because all subsequent information is based on the data provided in this phase, it is essential that the points are selected precisely. The computation phase of analysis is performed after all camera views have been digitized. The purpose of this phase is to compute the true threedimensional image space coordinates of the subject's body joints from the twodimensional digitized coordinates of each camera's view. Computation is performed using a direct linear transformation. This transformation is determined by first relating the known image space locations of the calibration points to the digitized coordinate locations of those points. The transformation is then applied to the digitized body joint locations to yield true image space locations. This process is performed under computer control with a small amount of timing information provided by the user. This information includes starting and ending points if all the data are not to be useds as well as a frame rate for the image sequence that may differ from the frame rates of the cameras used to record the sequence. When transformation is complete, a smoothing or filtering operation is performed on the image coordinates to remove small random digitizing errors and to compute body joint velocities and accelerations. Smoothing options include cubic and quintic splines as well as a Butterworth 2nd order digital filter (14,16,19). Smoothing may be performed automatically by the computer or interactively with the user controlling the amount of smoothing applied to each joint. In addition, error measurements from the digitizing phase may be used to optimize the amount of smoothing selected. In those cases involving impact, such as hitting a golf-ball, the point of impact can be identified during the smoothing process. The operator can select that point and several points on either side of the event so that the smoothing algorithm will adjust to this violent activity in an appropriate manner. That is to say, the motion before and after impact as well as impact itself can be more accurately calculated during the smoothing process. At the completion of smoothing the true threedimensional body joint displacements, velocities and accelerations have been computed on a continuous basis throughout the duration of the sequence. At this point, optional kinetic calculations may be performed to complete the computation phase. Body joint displacements, velocities and accelerations are combined with body segment mass distribution to compute dynamic forces and moment at each of the body joints. Muscular contribution to these forces and moments can then be computed by selectively removing the inertial and gravitational kinetic components. The presentation phase of analysis allows computed results to be viewed and recorded in a number of different formats. Body position and body motion can be presented in both still frame and animated "stick figure" format in three dimensions. Multiple .stick figures may be displayed simultaneously for comparison purposes. Joint velocity and acceleration vectors may be added to the stick figures to show the magnitude and direction of body motion parameters. Color hard copies of these displays can also be produced for reporting and publication. Results can also be reported graphically. Plots of body joint and segment linear and angular displacements, velocities, accelerations, forces and moments can be produced in a number of format options. An interactive graphically oriented user interface makes the selection and plotting of such results simple and straightforward. In addition, results may also be reported in numerical form. All quantities that can be selected for graphing may also be printed in tables of body motion parameters. The reliablity of the system are high with reliability coefficients ranging from .95 to .99. Coefficient variability is affected by poor and/or inappropriate video quality, inexact calibration points, and the skill level of the digitizing operators. Consider, for example a hypothetical situation. The civilian authorities of A City must describe, evaluate, promote or demote, train, and, perhaps, rehabilitating various professionals. These professions could include fire fighters, police personnel, sanitation workers, bus drivers, postal workers, etc. Modern societal demands insist that job related decisions must not discriminate but no effectively objective, accurate measurement devices have been available for the personnel selection process. How then could these authorities handle this hypothetical personnel situation? Imagine the following scenario. Fire fighters from A City choose a panel of their peers who select certain tasks that are the most frequently demanded in their jobs. These tasks are unanimously agreed upon and some specifically selected individuals, whom all fire fighters agree are superior examples of their profession, are chosen for testing. Biomechanical data is obtained on all those individuals as they perform each task. This data is processed to quantify a "generic" fire fighter so that guidelines for task performance can be described in precise, accurate, and duplicable form. Performance levels may be required for hiring, promotion, alternative assignments, and/or termination. The critical factor, however, is that biomechanical quantification can provide exact results without bias. The same sequential procedures could be utilized for testing the efficacy of a drug. For example, what is the difference between a type of operation and a new drug for patients afflicted with Parkinson's disease? Individuals could be analyzed biomechanically before and after the applications. In addition, evaluation of their fitness relative to their abilities to perform tasks could be monitored in terms of patterns of motion as well as strength. Physical therapy (15,18) and job retraining are other fields which provide numerous opportunities for the quantification of performance. ..If data has been recorded on the individual before an injury, comparison can easily be made to determinine level of recovery and/or disability. If an adequate data base has been created, it is also possible to compare an individual's activity with an "ideal" or "average" performance level. Other obvious examples for quantification of performance can be found in sports 13,17). Specific performance parameter measurements for various occupations coupled with actual performance analysis of individual's will allow better job description, work related performance criteria, and rehabilitation programs to be established. Biomechanical analyses and computer technology have increased the parameters upon which performance can be considered. Biomechanics provides a precise, objective, and quantifiable criterion for performance evaluation. Obviously, data about the arteries, the functioning capacities of various internal organs, and other anatomical or physiological data are not provided by biomechanical testing. However, the mare blind men employed to touch the elephant, the greater the chance to know what the elephant actually is. However, as the definitions and the evaluative instruments become more precise, it can only be hoped that the blind will be able to see. REFERENCES 1. G.B. Ariel, "Computerized Biomechanical Analysis of Human Performance," Mechanics and Sport, Vol. 4, pp. 267-275, The American Society of Mechanical Engineers, New York, 1973. 2. G.B. Ariel, "The Effect of Knee Joint Angle on Harvard Step Test Performance," Ergonomics. Vol.12, pp. 33-37, 1969. 3. G.B. Ariel, "Computer Application of Biomechanical Analysis of Human Performance in Sport and Industry," Canadian Congress of Sport and Physical Activity. Abstracts, Oct., 1973. 4. G.B. Ariel, "Computerized Biomechanical Analysis of the Knee Joint during Deep Knee Bend with Heavy Load," Biomechanics IV. Edited by Richard C. Nelson and Chauncey A. Morehouse, Fourth International.Seminar on Biomechanics Volume IV. The Pennsylvania State University, 1973. 5. G.B. Ariel, "Shear and Compression Forces in the Knee Joint during Deep Knee Bends" XXth World Congress in Sports Medicine Handbook, Melbourne, Australia, 1974. 6. G.B. Ariel, "Computerized Biomechanical Analysis of Human Performance," XXth World Con ress..in Sports Medicine Handbook, Melbourne, Australia, 1974. 7. G.B. Ariel, "Method for Biomechanical Analysis of Human Performance," Research Quarterly, Vol. 45, pp. 72-79, 1974. 8. G.B. Ariel, "Biomechanical Consideration in the Design and Construction of Resistance Exercise Equipment," Proceedings of the 1st Annual Meeting, Canadian Society for Biomechanics. University of Alberta, Edmonton, Alberta, Canada, pp.25-34, 1974. 9. G.B. Ariel, "The Biomechanics of Athletic Shoe Designs" Medicine and Science in Sports, (Abstract), Vol. 7, pp. 78, 1975. 10. G.B. Ariel, "Computer Method of Analyzing Locomotive Patterns of Handicapped Children," Final Report of the State of the Art Research Review and Conference on the Psychomotor Development in Preschool Handicapped Children for the Bureau of Education. Prepared for the Bureau of Education for the Handicapped Office of Education, Department of Health, Education, and Welfare. Contract No., 300-75-0225. pp. 16-25, 1976. 11. G.B. Ariel, "Neural Control of Locomotion - A Kinetic Analysis of the Trot in Cats," In Neural Control of Locomotion. Edited by R.M. Herman, et al. pp. 759-762, Plenum Publishing Corp., 1976. 12. G.B. Ariel, "Human Movements Analysis," Applied Ergonomics, Vol. 11, pp. 61-62, 1980. 13. G.B. Ariel, "Biomechanics," In Scientific Foundations of Sports Medicine. Edited by Carol C. Teitz, M.D., Chapter 12, pp. 271-297, Toronto: B.C. Decker, Inc., 1989. 14. J.F. Kaiser, "Digital Filters," Digital Filters and the Fast Fourier Transform, Edited by D. Liu, pp. 5-79, Dowden, Hutchinson & Ross, Stroudsburg, 1975. 15. I. Llacera and R. Squires. "An Analysis of the Shoulder Musculature during the Forehand Racquetball Serve," presented at American Physical Therapy Association meeting, Las Vegas, June, 1988. 16. C. Reinsch, "Smoothing by Spline Functions," Numerische Mathematik, Vol. 10, pp. 177-183, 1967. 17. P. Susanka, "Biomechanical Analyses of Men's Handball," Internationale Handball Federation 12th Men's Handball World Championship, Charles University, Prague, Czechoslovakia, 1990. 18. R.W. Wainwright, R.R. Squires, R.A. Mustich, "Clinical Significance of Ground Reaction Forces in Rehabilitation and Sports Medicine," presented at the Canadian Society for Biomechanics, 5th Biannial Conference on Biomechanics and Symposium on Human Locomotion, 1988. 19. G.A. Wood and L.S. Jennings, "On the Use of Spline Functions for Data Smoothing," J. of Biomechanics. Vol. 12 (6?, pp. 477-479, 1975.
|
[Back to Personal] [Back to Homepage] |