A camera is placed such that it
records the side view of play i.e. the view
from square leg or point. From another
camera the top view of play, directly
above the pitch is recorded. From the
videos obtained from these two cameras
the position of the ball in consecutive
frames is obtained. Using the ball
position in consecutive frames a curve
representing the equation of motion of
the ball is obtained. If the ball hadn’t hit
the pads it would have continued in its
equation of motion. By extrapolating the
equation of motion of the ball to the
stumps it could be Play is continuously
recorded in this camera. Whenever a
close LBW appeal occurs the video of
that particular delivery is cut. The pixel
value of the ball in consecutive frames is
then obtained using a MATLAB
program. Since we know that any
projectile determined if it would have hit
the stumps.
Original Title
An Image Processing Technique For LBW Decision Making Based on 2 Dimensional Analysis
A camera is placed such that it
records the side view of play i.e. the view
from square leg or point. From another
camera the top view of play, directly
above the pitch is recorded. From the
videos obtained from these two cameras
the position of the ball in consecutive
frames is obtained. Using the ball
position in consecutive frames a curve
representing the equation of motion of
the ball is obtained. If the ball hadn’t hit
the pads it would have continued in its
equation of motion. By extrapolating the
equation of motion of the ball to the
stumps it could be Play is continuously
recorded in this camera. Whenever a
close LBW appeal occurs the video of
that particular delivery is cut. The pixel
value of the ball in consecutive frames is
then obtained using a MATLAB
program. Since we know that any
projectile determined if it would have hit
the stumps.
A camera is placed such that it
records the side view of play i.e. the view
from square leg or point. From another
camera the top view of play, directly
above the pitch is recorded. From the
videos obtained from these two cameras
the position of the ball in consecutive
frames is obtained. Using the ball
position in consecutive frames a curve
representing the equation of motion of
the ball is obtained. If the ball hadn’t hit
the pads it would have continued in its
equation of motion. By extrapolating the
equation of motion of the ball to the
stumps it could be Play is continuously
recorded in this camera. Whenever a
close LBW appeal occurs the video of
that particular delivery is cut. The pixel
value of the ball in consecutive frames is
then obtained using a MATLAB
program. Since we know that any
projectile determined if it would have hit
the stumps.
International Journal of Image Processing and Visual Communication
Volume 1 , Issue1, August 2012
24
An Image Processing Technique For LBW Decision Making Based on 2 Dimensional Analysis
K.Vikram Gajanan 1 and C.Udaya Kumar 1
1 BSA Crescent Engineering College viki_341@yahoo.com
Abstract A camera is placed such that it records the side view of play i.e. the view from square leg or point. From another camera the top view of play, directly above the pitch is recorded. From the videos obtained from these two cameras the position of the ball in consecutive frames is obtained. Using the ball position in consecutive frames a curve representing the equation of motion of the ball is obtained. If the ball hadnt hit the pads it would have continued in its equation of motion. By extrapolating the equation of motion of the ball to the stumps it could be Play is continuously recorded in this camera. Whenever a close LBW appeal occurs the video of that particular delivery is cut. The pixel value of the ball in consecutive frames is then obtained using a MATLAB program. Since we know that any projectile determined if it would have hit the stumps.
Introduction Now a days the hawk eye (ball tracking) technology is used to determine the correctness of LBW decisions. It is based on the 3 dimensional tracking of the ball. The image is digitized and in the process sampled and quantized and so the accuracy of prediction suffers. On the other hand the method proposed in this paper tracks the ball two dimensionally on two different planes wherein the digitization of the image is not required and so the method proposed here doesnt compromise on accuracy.
Principle of operation The camera is placed as shown in the figure below and it gives the side view of play.
projected from the earths surface moves in a parabolic path, the ball after pitching travels in a parabolic path when viewed on a 2 dimensional plane parallel to the plane of the side camera(x-y plane). From the pixel values of the ball a parabolic curve is fit. This curve gives the equation of motion of the ball. The equation of the parabola is of the form y = f(x) = ax^2 + bx + c=0. By the method of interpolation the values of the coefficients a and b can be obtained in MATLAB.Having obtained the equation of motion the ball we can say that if the ball hadnt hit the pads it would have continued in its equation of motion. Therefore by tracing the equation of motion of the ball to the stumps we can say if it would have gone to hit the stumps. To do this from any of the frames the pixel value of the top of the International Journal of Image Processing and Visual Communication Volume 1 , Issue1, August 2012 25
stump is noted (say hx, hy). Substitute hx for x in the parabolic equation and obtain the corresponding y value. This The other camera is placed as shown in the figure below. It gives the top view of play. Play is continuously recorded in this camera as well. Whenever a close LBW . value of y is compared with the value of hy. If this value of y is lesser than hy then it can be concluded that the ball is within the height of the stumps else the ball is above the height of the stumps.
In the figure below
the curve which traces the ball is the required parabolic equation. The value of x at the stumps hx is substituted in the parabolic equation. The obtained value of y is compared with the value of y at the top of the stumps (i.e.) hy. If y is lesser than hy it can be concluded that the ball is within the height of the stumps else the ball is above the height of the stumps. The other camera is placed as shown in the figure below. It gives the top view of play. Play is continuously recorded in this camera as well. Whenever a close LBW appeal occurs the video of that particular delivery is cut. The pixel value of the ball in consecutive frames is then obtained using a MATLAB program. From the various pixel values of the ball a parabolic curve is fit. This equation gives the equation of motion of the ball in a 2 dimensional plane parallel to the plane of the top camera(x-z plane). The equation of this parabolic curve is of the form z = f(x) = ax^2 + bx +c =0.
By the method of interpolation the values of the coefficients a and b are obtained using MATLAB. Having obtained the equation of motion of the ball we can say that if the ball hadnt hit the pads it would have continued in its equation of motion. Therefore by tracing the equation of the ball to the stumps we can say if it would have gone to hit the stumps. To do this from any of the frames the pixel values of the left extreme and right extreme of the stumps are noted (say (sx, sz1) and (sx, sz2)). Substitute sx for x in the above parabolic equation and obtain the corresponding z value. If the z value lies within sz1 and sz2 (i.e.) within the left and the right extremes of the off stump and the leg stump resp. then the ball will hit within the range of the stumps. If the z value doesnt lie within the left and right extremes of the stumps then the ball will hit outside the range of the stumps. International Journal of Image Processing and Visual Communication Volume 1 , Issue1, August 2012 26
From the top camera the frame corresponding to the instant at which the ball hits the batsmens pads is captured. The value of the z co-ordinate at this instant is compared with the z co- ordinate of the off stump or the leg stump, depending upon the delivery. If the ball hits the batsmens pads outside the range of the stumps i.e. if the point of impact is outside the stumps the batsmen can straight away be declared NOT OUT.
If the ball pitches outside the leg stump it can be determined in a similar way as explained above.
Hence from the above arguments it can be concluded that, if all the conditions of both the cameras evaluate to TRUE (i.e.) if the ball lies within the height of the stumps, the ball lies within the expanse of the stumps (within the leg stump and off stump), the point of impact is within the stumps and if the ball doesnt pitch outside the leg stump the ball will surely go on to hit the stumps and the batsmen is declared OUT. If even any one of the conditions evaluates to false the ball wont hit the stumps and the batsmen is NOT OUT. This technology is so simple that it can be implemented in television sets. For customers using television services from DTH service providers, they can demand video coverage from any of the camera. Hence when a spectator watching a cricket match from his home has any doubt if it was an LBW he can demand the top camera and the side camera videos of the previous ball. The demand of the spectator is accepted and the top and the side camera videos of the previous ball is displayed on the television. During this time a video recorder when attached to the television records these videos. The videos thus recorded are processed as described above. This processing can be accomplished even by means of a low end processor which can be built within the television as a new appendage.
A software can be developed which obtains the pixel information of the ball from those samples and uses these information to predict whether the ball would have hit the stumps.
Advantages of this technology The present hawk eye technology that is being used tracks the ball 3 dimensionally. The image here, is digitized and in the process sampled and quantized and so the accuracy of prediction is reduced. On the other hand the method proposed in this paper tracks the ball two dimensionally. No sampling or quantization is employed and thus the method proposed here doesnt compromise on accuracy. Moreover this method is very simple to implement. A high degree of accuracy is achieved at a very low cost.
At present a minimum of 3 cameras are required to locate a point in space. But in this technique we are able to track the motion of the ball with just two cameras which is an advancement in technology. International Journal of Image Processing and Visual Communication Volume 1 , Issue1, August 2012 27
Limitations Umpires or some fielders may restrict the view of the ball from the camera. In this case the view of the ball is blocked by the fielder for an instant of time. Hence we wont be able to obtain the pixel information of the ball during that instant. This may reduce the accuracy of prediction.
Real time problems encountered When the sampled video signals are projected two dimensionally on the computer screen the pixel value of the pitch at different places will be different. This implies that the y co-ordinate of the
Another problem is the fixing of top camera. As seen in the figure above the top camera which gives the top view is fixed to a rod that runs along the diameter of the cricket field. But now a days in ICL cricket matches such type of cameras fixed in such positions are used. Further the position of the camera on the rod is controlled by a remote control. pitch is different at different places (consider the side view). But in reality the whole pitch is on the ground and hence it should have only one unique value as its y co-ordinate. To solve this problem a simple solution would be to place the side camera at the same level as the pitch (i.e.) on the ground. In this case the pitch would be seen as a straight horizontal line and the stump would be seen as a straight vertical line. Also the base of the camera should be placed parallel to the pitch. When the camera is placed in this way the edge of the pitch has a single unique value as its y co-ordinate.
Conclusion From the idea presented in this paper it can be concluded that, this idea would yield higher accuracy as compared to the present hawk eye system. At the same time it is much cheaper than the hawk eye system. Implementing this idea is not difficult. Simple MATLAB coding will do.