Combine video frames to calculate average motion of a plume

Hi,

For my master these we make sediment plumes (like smoke plumes, but with sediment) in water and study the motion of the plume.

We film it and have to calculate the average motion, such as the boundaries where the plume goes and the centerline of the plume.

So what I would like to do, is, from all frames, calculate the average plume...

Is there somebody with a good idea?

Thank you!

Stijn

Answer 1

[Image Analyst]

Average plume? You mean average image, since there is a plume in every image? Sum them up and divide by the number of images??? Make sure you cast to single or double first because summing uint8's will clip to 255.

[StijnBr]

Thank you for the quick response. I forgot to cast to single our double.

Now we want to calculate the dimensions of the plume.

Is it possible to add an x-axis and an y-axis on the picture starting in a certain point of the picture and plot the theoretical axisline of the plume and the experimental axisline of the plume?

Any hints to solve this problem? I found only information where the x-axis and y-axis start in a corner of the picture...

Regards,

Stijn

[Image Analyst]

See my BlobsDemo in my File Exchange for a tutuorial on how to make measurements on the image using regionprops. http://www.mathworks.com/matlabcentral/fileexchange/?term=authorid%3A31862

Answer 2

[James morgenstern]

you need to process each image separately. i would segment the plume from the background. Now extract geometric measures from the plume: get the centroid, calculate the spatial covariance 2d matrix. then the covariance matrix can be analyzed, for example get the eigen vectors which approximate the plume as an ellipse and give you the orientation and the length of the 2 axes.

Answer 3

[James morgenstern]

it can be dangerous to simply sum the images. does not account for any drift in the centroid of the plume! you need a process to account for the dynamics of the plume.