how to identify one impact-rebound trajectory from an image?

I'm going to identify the velocities of a particle before and after it impact to a flat surface. The images will be obtained from a high-frame camera, and the trajectories of particles can be recorded as streak lines in the image. Since there many particles' trajectories on the image, I want to use codes to identify and highlight the preferred impact-rebound trajectory and determine the angles between the wall and the distance between two closest centre points (shown in the figure). Each white line is represented using linear regression as a 1st order function, and the slope of each line can be obtained.

Based on literature, the criteria to select two streak lines before and after impact are:

1. The slope of the regression lines should be almost identical.
2. The intersect points of two streak lines with the surface line should be very close.
3. The ratio of streak line length A to the streak line separation B should be within certain limits.

I managed to represents all lines as linear functions and identified the slope of each function. And what I'm trying to do now is to identify each streak line from lines with a close slope value. I stucked here for several days and has no idea of how to do that. Any idea or advice will be appreciated.

Answer 1

[Image Analyst]

Since there is so much clutter in the image my suggestion is to ask the user to manually identify the streaks that you want using ginput() and bwselect() or ismember(). This will be a whole lot faster than spending days trying to do it automatically.

[Shipu HAN]

Thanks for your suggestion. But the problem is that I need to deal with many images rather than just one, in that case I'm not sure if it will be faster to manually identify the streak...

[Image Analyst]

To process a sequence of files, See the FAQ

[Shipu HAN]

Thanks a lot.

[Image Analyst]

My only suggestion to identify angled streaks is to use regionprops() and ask for orientation, and look for blobs that have an orientation in the angular range you expect.

[Shipu HAN]

Thanks, I will have a try.