PIV找不到particle

monkeyhellen

我们20米的方水池中，造波，然后用2.5cmby 2.5cm的塑料tracker来追踪流场，得到的图像像素比是2PIXEL/CM,frequency is 10 frame/sec, so the maximum offset between two image would be 6 pixel, and one particle is about 5*5 pixel in the image. 我用了很多interrogation length, and search length, 始终得不到流场，our PIV program works well under other application.

monkeyhellen

if anyone has a idea, please do say something. my image is too large to upload.

swell2005

Then you may upload any part of your image.But what do you mean to show the picture upstairs?

yxjbuaa

存在的问题可能是图像质量太差，将图像作后适当增强处理后可能可以得到结果。你传上来的图中的黑点是粒子吗？如果是可以将图像反相。
此外，根据你给的数据，建议你将粒子适当减小或者将测量区域增大，粒子图像的直径在2-3pixels较为理想，否则精度会降低，同时尽量将粒子位移增大（调节dt时间）也是需要的。

swell2005

请问一般情况下粒子位移是多少pixels？

yxjbuaa

粒子位移当然是越大越好，越大精度越高，但是需要同时考虑分辨率的影响。也就是说，在你希望得到的分辨率下（主要由查问域的大小来确定）粒子位移应当尽量的大，在使用了窗口平移技术后，粒子位移可以超过1/2查问域宽度。可以这样估算一下，当粒子图像的直径能够保证在2-3pixels时，粒子位移的最小偏差可以达到0.1pixels左右，那么如果你的粒子位移是10pixels，那么误差就至少有1%。注意，以上的这些推算只有理论指导意义，但是可以作为试验参数设定的一些必要参考准则。

monkeyhellen

the particle has the 5 by 5 pixel size, but because of the maneuverablity, the density is not uniform. I have tried inverse color method, it is not helpful. The maximum offset of the particle here is not more than 6 pixel between per pair of frames, normally only 2-3 pixel offset.Like I have said the sampling frequency of image is 10 frames per second, is it too fast? I need the instantaneous velocity to this interval.
To yxjbuaa: what do you mean by increasing the measurement area? that is  the area we are interested, and the particles are all what we have. Under this situation, do you think the only option left is to increase the sampling interval to obtain bigger offset?
here is the image with inhanced particles:

monkeyhellen

this processed image is little bit different from my raw image, because the of the contrast between particles and contrast. the particles in this image add to about 10 pixel size. Has anybody have an idea what interrogation size should I use for this?

monkeyhellen

robin

如果你能把图像发给我的话。我可以用我的软件作一下预处理看一下效果如何。