c******r
发帖数: 132 | 1 看见有人问，不记得以前发过没有了。再发一次，温习一下吧。
一般录像的帧数是30fps（帧/秒）。通过录像数帧数，从球触拍开始到触地为止。比如
说是20帧，对应0.667秒。然后估算球飞行的距离。假定是直线，用勾股定理就可以搞
定了。发T角的话，从底线到对面的发球线60‘（18.3m)，加上身高+手+拍大约3m，总
共18.5m飞行距离。18.5/0.667秒=28m/s=100mph.
附：可以数帧数的播放软件，推荐kmplayer或射手影音播放器。
以上方法经本人测试，结果合理。测试是一段费费的比赛录像，15帧，120mph。
好久没写实验报告了，写饿了，有木有包子吃啊。 |
K****D
发帖数: 30533 | 2 The problem with this method is that it's difficult to decide the exact
frame where the ball touches the racket or the ground.
My fastest serve caught on camera was about 70-80mph.
Also your calculation is incorrect. 28 m/s = 100 kmph, not 100 mph. Federer'
s
video is definately lower than 15 frames.
【在 c******r 的大作中提到】
: 看见有人问，不记得以前发过没有了。再发一次，温习一下吧。
: 一般录像的帧数是30fps（帧/秒）。通过录像数帧数，从球触拍开始到触地为止。比如
: 说是20帧，对应0.667秒。然后估算球飞行的距离。假定是直线，用勾股定理就可以搞
: 定了。发T角的话，从底线到对面的发球线60‘（18.3m)，加上身高+手+拍大约3m，总
: 共18.5m飞行距离。18.5/0.667秒=28m/s=100mph.
: 附：可以数帧数的播放软件，推荐kmplayer或射手影音播放器。
: 以上方法经本人测试，结果合理。测试是一段费费的比赛录像，15帧，120mph。
: 好久没写实验报告了，写饿了，有木有包子吃啊。
|
a*****0
发帖数: 6788 | 3 Wow, the OP's one of the most useful posts I've seen here.
The basic principle is sound. All you need is a high-speed camera, e.g. one
of those cheap Casio cameras that can shoot 120, 210, 1000/fps, to get this
more precisely. You can further calibrate the method by shooting some
other moving targets, like yourself running 100 meters on a track, which you
can time with a stop watch, then compare to the camera to see how good the
measuring is.
I may go try it this weekend :)
Federer's: video is definately lower than 15 frames.
【在 K****D 的大作中提到】
: The problem with this method is that it's difficult to decide the exact
: frame where the ball touches the racket or the ground.
: My fastest serve caught on camera was about 70-80mph.
: Also your calculation is incorrect. 28 m/s = 100 kmph, not 100 mph. Federer'
: s
: video is definately lower than 15 frames.
|
a**s
发帖数: 9606 | 4 His calculation is the average speed. The serve speed is claculated
as the speed leaving the racquet. No way you can calculate that from
video recording.
Federer'
【在 K****D 的大作中提到】
: The problem with this method is that it's difficult to decide the exact
: frame where the ball touches the racket or the ground.
: My fastest serve caught on camera was about 70-80mph.
: Also your calculation is incorrect. 28 m/s = 100 kmph, not 100 mph. Federer'
: s
: video is definately lower than 15 frames.
|
a*****0
发帖数: 6788 | 5
A 1000fps video recording the first moment of impact can make this to work:
For example, 100mph serve is 44.7 meters/second. At 1000fps, it's 4.47 centimeter per frame. Using the racket length as a reference, you can take the first 20 frames (44.7 centimeters) from when the racket hits the ball to measure the speed. If a serve can go the full length of a standard racket (27in=68.6cm) in 20 frames, it's 153mph! assuming the deceleration in such a small distance is negligible.
【在 a**s 的大作中提到】
: His calculation is the average speed. The serve speed is claculated
: as the speed leaving the racquet. No way you can calculate that from
: video recording.
:
: Federer'
|
j*******j
发帖数: 150 | 6 This makes sense!
Actually, the eagle eye system consists of three high speed cameras, and the
3D is reconstructed from the 2D images.
centimeter per frame. Using the racket length as a reference, you can take
the first 20 frames (44.7 centimeters) from when the racket hits the ball to
measure the speed. If a serve can go the full length of a standard racket
(27in=68.6cm) in 20 frames, it's 153mph! assuming the deceleration in such a
small distance is negligible.
【在 a*****0 的大作中提到】
:
: A 1000fps video recording the first moment of impact can make this to work:
: For example, 100mph serve is 44.7 meters/second. At 1000fps, it's 4.47 centimeter per frame. Using the racket length as a reference, you can take the first 20 frames (44.7 centimeters) from when the racket hits the ball to measure the speed. If a serve can go the full length of a standard racket (27in=68.6cm) in 20 frames, it's 153mph! assuming the deceleration in such a small distance is negligible.
|
K****D
发帖数: 30533 | 7 Those two speeds actually don't differ much.
It's after the first bounce, that the ball loses quite some speed.
【在 a**s 的大作中提到】
: His calculation is the average speed. The serve speed is claculated
: as the speed leaving the racquet. No way you can calculate that from
: video recording.
:
: Federer'
|
b*****n
发帖数: 17570 | 8 帧数怎么数？
【在 c******r 的大作中提到】
: 看见有人问，不记得以前发过没有了。再发一次，温习一下吧。
: 一般录像的帧数是30fps（帧/秒）。通过录像数帧数，从球触拍开始到触地为止。比如
: 说是20帧，对应0.667秒。然后估算球飞行的距离。假定是直线，用勾股定理就可以搞
: 定了。发T角的话，从底线到对面的发球线60‘（18.3m)，加上身高+手+拍大约3m，总
: 共18.5m飞行距离。18.5/0.667秒=28m/s=100mph.
: 附：可以数帧数的播放软件，推荐kmplayer或射手影音播放器。
: 以上方法经本人测试，结果合理。测试是一段费费的比赛录像，15帧，120mph。
: 好久没写实验报告了，写饿了，有木有包子吃啊。
|
a*****0
发帖数: 6788 | 9
In addition to the players mentioned in OP, the best is to use a video
editing tool, zoom in on the timeline.
【在 b*****n 的大作中提到】
: 帧数怎么数？
|
K****D
发帖数: 30533 | 10 You made a fundamental error.
The trajectory of the ball is random. It can be parallel to the
lens, can be vertical. Using your method, you will get anything
from 0-153mph, hehe.
centimeter per frame. Using the racket length as a reference, you can take
the first 20 frames (44.7 centimeters) from when the racket hits the ball to
measure the speed. If a serve can go the full length of a standard racket
(27in=68.6cm) in 20 frames, it's 153mph! assuming the deceleration in such a
small distance is negligible.
【在 a*****0 的大作中提到】
:
: In addition to the players mentioned in OP, the best is to use a video
: editing tool, zoom in on the timeline.
|
|
|
a*****0
发帖数: 6788 | 11 Dude, if the camera is placed on the side along the baseline, and you are unable to serve down the T (which means a near-90-degree angle with the baseline), you shouldn't bother with knowing how fast you serve.
【在 K****D 的大作中提到】
: You made a fundamental error.
: The trajectory of the ball is random. It can be parallel to the
: lens, can be vertical. Using your method, you will get anything
: from 0-153mph, hehe.
:
: centimeter per frame. Using the racket length as a reference, you can take
: the first 20 frames (44.7 centimeters) from when the racket hits the ball to
: measure the speed. If a serve can go the full length of a standard racket
: (27in=68.6cm) in 20 frames, it's 153mph! assuming the deceleration in such a
: small distance is negligible.
|
c******r
发帖数: 132 | 12
I take this as a compliment. :)
one
this
you
the
【在 a*****0 的大作中提到】
: Wow, the OP's one of the most useful posts I've seen here.
: The basic principle is sound. All you need is a high-speed camera, e.g. one
: of those cheap Casio cameras that can shoot 120, 210, 1000/fps, to get this
: more precisely. You can further calibrate the method by shooting some
: other moving targets, like yourself running 100 meters on a track, which you
: can time with a stop watch, then compare to the camera to see how good the
: measuring is.
: I may go try it this weekend :)
:
: Federer's: video is definately lower than 15 frames.
|
c******r
发帖数: 132 | 13 data king is correct on this. Federer's serve should be about 10 frames.
Federer'
【在 K****D 的大作中提到】
: The problem with this method is that it's difficult to decide the exact
: frame where the ball touches the racket or the ground.
: My fastest serve caught on camera was about 70-80mph.
: Also your calculation is incorrect. 28 m/s = 100 kmph, not 100 mph. Federer'
: s
: video is definately lower than 15 frames.
|
c******r
发帖数: 132 | 14 The average speed before the ball bounce is a correct measure for the serve
speed. It's what the receiving player sees. It doesn't make sense to measure
the instantaneous velocity you mentioned. It has no practical use and it's
not the speed gun measured in matches, which is an average speed while the
ball is in air traveling, depending on where you put the gun.
【在 a**s 的大作中提到】
: His calculation is the average speed. The serve speed is claculated
: as the speed leaving the racquet. No way you can calculate that from
: video recording.
:
: Federer'
|
s*****k
发帖数: 1630 | |
K****D
发帖数: 30533 | 16 Nope. In pro matches, the radar gun does record the instantaneous
velocity. It is placed behind the server.
However, as I mentioned, the average speed and the instantaneous
speed doesn't differ much. The ball loses at least 30% speed after
first bounce though.
serve
measure
s
【在 c******r 的大作中提到】
: The average speed before the ball bounce is a correct measure for the serve
: speed. It's what the receiving player sees. It doesn't make sense to measure
: the instantaneous velocity you mentioned. It has no practical use and it's
: not the speed gun measured in matches, which is an average speed while the
: ball is in air traveling, depending on where you put the gun.
|
K****D
发帖数: 30533 | 17 Your serve in the ben is probably in the 50-60mph range.
【在 s*****k 的大作中提到】
: 等你们研究出结果了偶再试。
|
s*****k
发帖数: 1630 | 18 误差多少？ 偶是该高兴还是不高兴？
【在 K****D 的大作中提到】
: Your serve in the ben is probably in the 50-60mph range.
|
K****D
发帖数: 30533 | 19 As stated, error is +/- 5mph.
I think you should have mixed feeling since you serve slower than
a 3.0 male like me, hehe.
【在 s*****k 的大作中提到】
: 误差多少？ 偶是该高兴还是不高兴？
|
a*m
发帖数: 6253 | 20 happy of coz.
That is very impressive, even faster than mine...
【在 s*****k 的大作中提到】
: 误差多少？ 偶是该高兴还是不高兴？
|
|
|
t*****i
发帖数: 2249 | 21 有的3。0男发球还是力量挺大的，但是不一定完全接不到。
落点更重要，50-60mph落点好非常有威胁
【在 K****D 的大作中提到】
: As stated, error is +/- 5mph.
: I think you should have mixed feeling since you serve slower than
: a 3.0 male like me, hehe.
|
t*****i
发帖数: 2249 | 22 老大你什么时候奔一个让小的们学习一下
【在 a*m 的大作中提到】
: happy of coz.
: That is very impressive, even faster than mine...
|
a**s
发帖数: 9606 | 23 ft! 咋这么谦虚了?
【在 a*m 的大作中提到】
: happy of coz.
: That is very impressive, even faster than mine...
|
a**s
发帖数: 9606 | 24 Just cheked Murray's 201km/h ace vs Nadal. Count 18-19 frames on
50fps 720p video. The pass is around 19m. So the caclulated speed
is 180-190km/h. So this method seems underestimated 5% to 10%.
【在 K****D 的大作中提到】
: Those two speeds actually don't differ much.
: It's after the first bounce, that the ball loses quite some speed.
|
K****D
发帖数: 30533 | 25 5%-10% is already pretty low error. As you mentioned, the error caused
by 1 frame count could easily be 5%-10% too (regular videos are 30fps).
I would worry about that first and wouldn't conclude the 5%-10% under-
estimation without extensive statistics.
In my experience, I often have trouble deciding the exact beginning
frame and the exact ending frame. I have to choose 1 frame out of
2 ambiguous ones.
Theoretically radar gun should always measure higher speed than the
video analysis method. But if the difference is just 5%, I'd say
the video analysis is a very good estimation.
【在 a**s 的大作中提到】
: Just cheked Murray's 201km/h ace vs Nadal. Count 18-19 frames on
: 50fps 720p video. The pass is around 19m. So the caclulated speed
: is 180-190km/h. So this method seems underestimated 5% to 10%.
|
a**s
发帖数: 9606 | 26 50fps will get 5-10% error, so regular 30fps will get >10% error.
Just want to point out that your statement about the average speed
= the init speed is not accurate.
【在 K****D 的大作中提到】
: 5%-10% is already pretty low error. As you mentioned, the error caused
: by 1 frame count could easily be 5%-10% too (regular videos are 30fps).
: I would worry about that first and wouldn't conclude the 5%-10% under-
: estimation without extensive statistics.
: In my experience, I often have trouble deciding the exact beginning
: frame and the exact ending frame. I have to choose 1 frame out of
: 2 ambiguous ones.
: Theoretically radar gun should always measure higher speed than the
: video analysis method. But if the difference is just 5%, I'd say
: the video analysis is a very good estimation.
|
a**s
发帖数: 9606 | 27 not that curious. Just want to check Kennedy's statement
a |
K****D
发帖数: 30533 | 28 I only said "they don't differ much" compared to the slow down after
the first bounce. I stated in a later post that the ball loses more
than 30% speed after first bounce.
Of course I know radar gun method is faster.
【在 a**s 的大作中提到】
: 50fps will get 5-10% error, so regular 30fps will get >10% error.
: Just want to point out that your statement about the average speed
: = the init speed is not accurate.
|
v***s
发帖数: 1893 | 29 you are wrong in this issue.
radar gun does measure the speed around the contact point. Have you seen 200
M/hr flying badminton ball? No, that is what radar gun measured badminton
speed. They don't measure the receiver feel. however, if you hit faster,
ball will fly faster.
lots of thing don't make sense. For instance, measuring car safty by driving
car to the concrete wall. It should measure the collision between a tested
car and a standard car.
serve
measure
s
【在 c******r 的大作中提到】
: The average speed before the ball bounce is a correct measure for the serve
: speed. It's what the receiving player sees. It doesn't make sense to measure
: the instantaneous velocity you mentioned. It has no practical use and it's
: not the speed gun measured in matches, which is an average speed while the
: ball is in air traveling, depending on where you put the gun.
|
