Content uploaded by David L Neumann
Author content
All content in this area was uploaded by David L Neumann
Content may be subject to copyright.
B370 1
Running head: EVALUATING PUTTING ACCURACY
A camera-based scoring system for evaluating performance accuracy during a golf
putting task
David L. Neumann1 & Patrick R. Thomas2
1 School of Psychology, Griffith University and Centre of Excellence for Applied
Sport Science Research, Queensland Academy of Sport
2 School of Education and Professional Studies, Griffith University and Centre of
Excellence for Applied Sport Science Research, Queensland Academy of Sport
Corresponding author and address: David Neumann, School of Psychology, Griffith
University (Gold Coast Campus), Mail: GRIFFITH UNIVERSITY QLD, Queensland,
4222, Australia, E-mail D.Neumann@griffith.edu.au, Facsimile +61(0)7 5552 8291,
Telephone +61(0)7 5552 8052
B370 2
Abstract
Golf putting accuracy is often evaluated by measuring the distance that the ball
finishes from the hole. However, accuracy is a function of line and length and
distance-from-hole measures confound these two factors. A scoring system for
evaluating putting accuracy is described that enables the efficient measurement of
errors in line and length. A camera placed above the hole takes digital photographs of
the final position of the ball. A custom developed program written in the National
Instruments LabView graphical programming language derives a variety of accuracy
measures from these photographs, including distance from hole, angle of error,
distance short or long from hole, and distance left or right from hole. Evaluation of
the system indicated that the measures were as accurate as manual measurements and
were reliable when rescored on separate occasions. The camera-based scoring system
presents a number of advantages in the evaluation of putting accuracy and may be
extended to examine performance in other sports.
Keywords: golf, putting, accuracy, performance, camera
B370 3
Introduction
Performance in sports that require an accurate aim like archery and rifle
shooting can be evaluated by measuring how far the shot finishes from the target.
However, for sports such as basketball, tennis, and football, targets can be missed
because of errors in the aim or errors in the force of the shot. In a similar vein, Pelz
(2002) identified two requirements for good putting in golf – the accuracy of the
initial line (aim) and the precision of rolling the ball at the proper speed. The
combination of these two factors determines whether the ball finishes in the hole. He
argued that aim, also referred to as direction, is a critical aspect of putting, with most
golfers substantially underestimating the true break of a putt. The golfer must also
anticipate how the speed and contours of the putting surface relate to the force that
needs to be imparted on the ball (Fairweather, Button, & Rae, 2002). Performance
measures should therefore be sensitive to both of these variables.
The most commonly used measure in the evaluation of golf stroke accuracy is
the distance the ball finishes from the target (e.g., Beilock & Carr, 2001; Beilock,
Carr, MacMahon, & Starkes, 2002; Crews & Landers, 1993). An alternative method
is to draw concentric circles of different radii around the target and award
increasingly higher points to balls finishing in zones closer to the target (e.g.,
Guadagnoli & Holcomb, 1999; Wulf, Lauterbach & Toole, 1999). However, neither
of these approaches shows whether the ball missed the target because of an error in
direction and/or an error in the force of the shot. Such combined measures may be
less sensitive to the influence of psychological factors (e.g., anxiety, attentional focus
strategy) on the different aspects of performance if these factors influence one aspect
of performance and not another. A scoring system that is capable of deriving
B370 4
performance measures sensitive to line and length thus has the potential to advance
the methods currently used in sport psychology research.
The present report describes the development and evaluation of a new method
to measure putting performance. The method comprises four steps. First, an
apparatus that can take digital photographs of the area around the target (e.g., hole) is
set up. Second, calibration of the system to enable the performance measures to be
translated into real measurement units (e.g., centimeters, degrees of error) is
completed. Third, photographs are taken of the final position of the ball during the
testing of a research participant. Fourth, the digital photographs are scored with the
use of computer software to derive a range of performance measures based on the line
and length of the putt.
Apparatus used to obtain the digital photographs
The apparatus used to obtain the digital photographs is shown in Figure 1. It
consisted of a digital camera that is attached to a tripod and boom. The digital camera
was a Sony Mini DV Handycam (Model DCRHC42) fitted with a 25 mm 0.6x Sony
Wide Conversion Lens (Model VCL-0625S). The camera can take moving videos or
still digital photographs, although only the still photographic function was used in this
application. It has a number of configurable settings and of these, the highest
resolution (1152 x 864 pixels), the widest lens aperture (i.e., no zoom), and the
date/time function were employed. The camera runs on its own battery power,
enabling the system to be used in locations without an accessible power supply. The
wide conversion lens was attached to the camera to ensure that a sufficiently large
area around the target could be photographed.
B370 5
-------------------------------
Insert Figure 1 about here
-------------------------------
The camera was attached to a tripod and boom. The tripod was a standard
metal tripod used for photography and was 1.42 m in height when erect. Attached to
the tripod was a telescopic carbon fiber boom pole that could be extended from 0.90
m to 3.9 m. The camera was attached to the end of the boom and due to its weight
(470 g, including battery) it was necessary to stabilize the tripod with small shot bags
to avoid the apparatus tipping over. The tripod and boom allowed the camera to be
suspended 3.4 m above the target hole. Suspending the camera in excess of this
height would allow the photographs to capture a wider area around the target hole.
However, we found that the resulting area coverage of approximately 5 m x 3.75 m
was adequate to capture all but the most erroneous putts when novice, experienced,
and elite athletes made a series of 20 putts 2.4 m from the hole.
System calibration
It is necessary to calibrate the system prior to recording a participant’s
performance. The first step in the calibration is to ensure that the camera lens is
suspended parallel to the level ground around the target hole. A circular bubble level
attached to the camera allowed this to be done. The second step is to take a
photograph of calibration objects to provide references for later scoring. The two
calibration objects are highlighted in the photograph in Figure 2. One of these objects
is a 1 m bar that can be placed at any location in the photograph. The second object is
a rope that is pulled straight along a line running from the center of the target hole to
the origin of the putt. In most cases, the putt origin will lie outside the photographed
B370 6
area. The calibration objects provide a means to measure distances and angle of error
in real units.
-------------------------------
Insert Figure 2 about here
-------------------------------
Taking photographs during testing of a research participant
The testing of research participants is efficient and proceeds quickly. All that
is required is that a photograph of the final position of the ball is taken after each putt.
The camera is supplied with a remote control to allow the photographs to be taken
even though the camera is suspended. Upon pressing the appropriate button on the
remote control, the camera emits an audible sound to provide verification that a
photograph has been taken. After taking the photographs, the ball can be removed
from the green and the participant can proceed with their next attempt. All
photographs are stored on a memory stick in the camera in JPEG format. The
photographs can be saved in an arbitrary name (e.g., DSC04561.JPG) that increases in
count with each successive photograph. The use of the date and time setting on the
camera provides a second means by which the data can be tracked back to a particular
participant.
Computer software used for scoring
Putting performance was scored through the use of a custom-developed
computer program written within the National Instruments LabVIEW programming
environment. The Laboratory Virtual Instrumentation Engineering Workbench
(LabVIEW) provides a graphical interface to develop custom-written programs that
can be ported to a variety of platforms, including Windows, Mac OS, and Linux. This
programming environment has been used to develop other data scoring and analysis
B370 7
programs in behavior research (e.g., Angrilli, 1995; Duley, Janelle, & Coombes,
2004). Programming within LabView involves interacting with a front panel that
provides a means to develop the graphic user interface (GUI) and a diagram panel that
provides a means to set the parameters and flow of data through the program.
Programs written in LabView are termed virtual instruments or VIs. The custom-
developed VI used to evaluate putting accuracy in this project was called
ScorePutting and contained a number of sub-VIs that used algorithms for calculating
the various performance parameters.
The GUI for ScorePutting is shown in Figure 2. The majority of the screen is
dedicated to displaying the photographs. The right panel of the screen contains
several controls and indicators to assist with scoring. Two controls that must be set
before starting the program are the Number of conditions and Distance of putt (m).
By specifying the number of conditions, the user can employ the same calibration
values and saved data file to score multiple experimental conditions from the same
participant. This might occur, for example, if the participant used different
performance routines in different blocks of putts. The distance of the putt needs to be
specified because the origin of the putt will rarely be shown in the photograph. The
final parameters that can be set by the user are the Default values for “not in photo”
putts. This is to be used for those rare occasions when a putt has been particularly
poor such that it does not lie within the photographed area. In these cases, the
experimenter should code the putt according to the direction in which the putt finished
using the coding sheet shown in the top right hand corner of the GUI. This will allow
estimate values or missing values, according to the user’s preference, to be used. The
estimate values can be calculated manually for the 12 different coded locations. The
B370 8
exact values will vary according to the individual set up because it will depend on the
coverage of the photographed area.
The sequence of operations that occurs upon starting ScorePutting is shown in
Figure 3. The program initially prompts for the filename to save the data. Identifying
characteristics of the participant (name/code and handedness) are requested and saved
in the data file. The latter characteristic may also be used to determine if shots are
missed by a push/slice or pull/hook, if desired. The JPEG file for the calibration
photograph is next displayed. The calibration proceeds by (a) clicking on the end
points of the 1 m calibration bar placed in the photo, (b) clicking on two points along
the straight line that runs from the center of the hole to the origin of the putt, and (c)
clicking on the center of the hole. A dot appears on the photograph for visual
verification during this process and there is an option to repeat the calibration if an
error is made. After the calibration values have been accepted, the performance data
may be scored. All that is required is to browse to the folder that contains the JPEG
photographs for the condition to be scored and select the folder. The program will
automatically load each photograph sequentially in alphanumerical order for scoring.
If more than one condition has been specified to be scored, the user will be prompted
for the folder that contains the photographs for the next condition and the procedure
repeats.
-------------------------------
Insert Figure 3 about here
-------------------------------
The scoring of the photographs proceeds quickly after calibration. If the ball
finished in the hole, the user clicks on the button “PUTT IN HOLE” and values are
automatically recorded (e.g., distance from hole = 0, angle of error = 0). If the ball
B370 9
finished outside of the photographed area, the user clicks on the button “PUTT NOT
IN PHOTO”. A dialogue box will be displayed prompting the user to specify the
numerical code for the appropriate estimated values. Finally, if the ball finished in the
photograph, but was not holed, the user simply clicks on the ball to measure
performance. In this case, the program will use a series of algorithms to calculate the
various parameters to measure performance. The performance measures that are
calculated are shown in Figure 4 and include (a) distance short or long, (b) length
error (distance of ball from origin minus distance from origin to hole), (c) angle of
error, (d) distance left or right, and (e) distance from hole. All except the angle of
error are expressed in centimeters and measures (a), (b), and (d) are expressed as
either positive or negative values to indicate the direction of the error (e.g., negative
values indicate that the final distance was short or left).
The first two measures (a, b) reflect accuracy in the force of the putt. In
addition to providing a measure of how well a golfer can read the speed of the green,
such measures may be sensitive to emotional factors (e.g., anxiety can increase
muscular tension and make it difficult to lag a putt effectively). The next two
measures (c, d) reflect accuracy in aim, identifying errors in direction as well as any
consistent patterns in those errors such as pushing or pulling putts. Separate
evaluation of distance and aiming errors will indicate whether a putt has missed
because of either or both of these factors. The final measure, distance from hole,
reflects a combination of distance and aim. It can provide a common metric for
comparisons with previous studies that did not separate distance and aiming in
evaluating performance (e.g., Beilock & Carr, 2001; Beilock et al., 2002; Crews &
Landers, 1993).
B370 10
-------------------------------
Insert Figure 4 about here
-------------------------------
The calculation of the various performance parameters is based on the
principle of dividing the photograph into x and y co-ordinates. A 1152 x 864 pixel
photograph thus consists of 1152 values along the x-axis and 864 values along the y-
axis, although the axes can be extended indefinitely to allow the position of objects
not in the photograph to be specified (e.g., the origin of the putt). The location of the
hole and the final position of the ball can thus be specified by separate x,y co-
ordinates and the line of the putt can be described by the use of a regression equation.
Using the various known x,y co-ordinates and the line of the putt, the methods of
trigonometry are applied to calculate the various performance parameters. In the case
of the distance measures, the values are originally calculated in pixels and
subsequently converted to centimeters. The conversion is possible because the
calibration of the system uses the 1 m bar to determine the ratio of how many pixels
correspond to 1 m. The final measure reflects the distance from the center of the ball
to the center of the relevant reference (e.g., hole). At the completion of scoring of all
photographs, all measured parameters are saved within a text formatted data file.
Evaluation of the system
To evaluate the accuracy and consistency of the system we took a sample of
20 photographs in which the ball was placed randomly from 12 cm to 210 cm in each
of the four quadrants around the hole. Photographs were taken at each ball position
and the distance of the ball from the hole was noted on a tape measure. When the
photographs were subsequently scored, the correlation between the computer
B370 11
calculated distance from hole and the manually measured distance from hole was very
high, r = 0.9996, p < .001. Test-retest reliability was also very high when the
photograph data were rescored one week later with a correlation between the two sets
of scores of r = 0.9999, p < .001.
Relative merits of the system
The camera-based scoring system presents a number of advantages to
encourage its use in the evaluation of putting performance. It provides a permanent
photographic record of the participant’s performance. This can be advantageous if
different performance measures need to be calculated from the same study at a later
date. The camera-based system considerably simplifies the testing procedure by
eliminating the need for the experimenter to make manual measurements. Apart from
the significant ergonomic advantages, the time taken to test a participant is
substantially reduced. The biggest strength of the system is, however, that it allows
several parameters of performance to be evaluated simultaneously. The common
measure of distance from hole is complemented by measures that separate
performance in line and length to determine if differences between groups of
participants or experimental conditions influence one or both performance factors. In
this way, it may be easier to identify systematic errors that occur during testing (e.g.,
pulling a ball when putting under pressure).
The camera-based scoring system also has a number of features that place
limitations on its use. When used outdoors, it can be influenced by weather
conditions in that strong winds can move the camera and it cannot be used when
raining. However, it would be unlikely that the system would be used under such
conditions since they may also hinder the participant’s performance. The camera,
while portable, is best used when all putts are directed at one hole as it would take
B370 12
some time to move the camera from one hole to another. Finally, the system was
developed and evaluated for putts taken on a flat, even surface. Some of the
performance measures, such as angle of error, may not be readily interpretable when
breaking putts are used.
Conclusions
The development of the camera-based scoring system has shown promise as a
means by which putting performance can be evaluated. The use of performance
measures that are sensitive to errors in line and length has the potential to enrich the
evaluation of a participant’s putting ability. There is potential for the general system
to be extended in other applications. For instance, the use of a video camera to
capture moving pictures might allow additional measures to be captured (e.g., pace)
or to evaluate performance during breaking putts. It may also be possible to apply the
same principles to evaluate performance during other aiming tasks that require
accuracy in line and length, such as basketball shooting. By capturing richer accuracy
measures it may be possible to provide a more fine-grained analysis of those
psychological and physical factors that influence performance during training or
competition.
B370 13
References
Angrilli, A. (1995). PSAAL: A LabVIEW 3 program for data acquisition and
analysis in psychophysiological experiments. Behavior, Research Methods,
Instruments & Computers, 27, 367-374.
Beilock, S. L., & Carr. T. H. (2001). On the fragility of skilled performance: What
governs choking under pressure? Journal of Experimental Psychology: General,
130, 701-725.
Beilock, S. L., Carr, T. H., MacMahon, C., & Starkes, J. L. (2002). When paying
attention becomes counterproductive: Impact of divided versus skill-focused
attention on novice and experienced performance of sensorimotor skills. Journal
of Experimental Psychology: Applied, 8, 6-16.
Duley, A. R., Janelle, C. M., & Coombes, S. A. (2004). An open-source LabVIEW
application toolkit for phasic heart rate analysis in psychophysiological research.
Behavior Research Methods, Instruments, & Computers, 36, 778-783.
Crews, D. J., & Landers, D. M. (1993). Electroencephalographic measures of
attentional patterns prior to the golf putt. Medicine and Science in Sports and
Exercise, 25, 116-126.
Fairweather, M.M., Button, C., & Rae, I. (2002). A critical examination of motor
control and transfer issues in putting. In E. Thain (Ed.), Science and golf IV:
Proceedings of the World Scientific Congress of Golf (pp. 100-112). London:
Routledge.
Guadagnoli, M., & Holcomb, W. (1999). Variable and constant practice: Ideas for
successful putting. In M.R. Farrally & A.J. Cochran (Eds.), Science and golf III:
Proceedings of the World Scientific Congress of Golf (pp. 261 – 270). Champaign,
IL: Human Kinetics.
B370 14
Pelz, D. (2002). Dave Pelz’s putting bible. London: Aurum.
Wulf, G., Lauterbach, B., & Toole, T. (1999). Learning advantages of an external
focus of attention in golf. Research Quarterly for Exercise and Sport, 70, 120-
126.
B370 15
Author Notes
This research was supported under Australian Research Council’s Linkage Projects
Funding Scheme (Project Number LP0667727). Thanks to Minhtri Pham and
Xuesong Lee for computer programming assistance and to Martin Hampson for
assistance during evaluation and testing.
B370 16
Appendix: Archived Materials
The following materials associated with this article may be accessed through
the Psychonomic Society’s Norms, Stimuli, and Data archive, found at
http://www.psychonomic.org/archive/. To access these files you may search for this
article by using the search parameters of the journal (Behavior Research Methods,
Instruments, & Computers), the first authors surname name (Neumann), and the
publication year (2007). There are two downloadable archives.
File: Scoreputting.zip
Description: This archive contains the Labview VI files required to run the
ScorePutting program. You must have LabView version 7.1 or higher installed on the
computer to run these files. The main file is called ScorePutting.vi and the subVIs
used by it are also included in the archive. It is possible to modify the LabView VI
files with the LabView program if necessary to customize the ScorePutting program
to meet individual needs.
File: InstallScorePutting.zip
Description: This archive contains a self-extracting archive of files. The files
include an .exe version of the ScorePutting program, the LabView runtime
environment, and application installer for MicroSoft Windows targets. Installation of
these files will allow the ScorePutting program to run on computers in which
LabView program is not installed. However, it is not possible to modify the LabView
VI files which were used to compile the executable file from.
B370 17
Figure Captions
Figure 1. A schematic of the set up for the apparatus used to take photographs of the
area around the target hole.
Figure 2. The graphic user interface for the ScorePutting program written in the
National Instruments LabVIEW programming language. The majority of the
interface shows the photograph of the area around the target hole. The photograph in
this figure shows the calibration objects of the 1 m bar and the rope. The interface
also contains a number of controls and indicators in the right panel that are used
during scoring.
Figure 3. A flow chart showing the sequence of operations that occur to evaluate
putting performance with the ScorePutting program.
Figure 4. The five performance measures derived from the photographs when the ball
did not finish in the hole.
B370 18
B370 19
B370 20
B370 21
