Is there a simple method for Pattern Recognition?

I am trying to write a script that recognises shapes and patterns within an image based on training data.

I'm unfamiliar with machine learning and deep learning, and the examples I've found online have been very confusing.

I have the following images

I'm going to crop each of those columns of water and use a few of them as training data. I need to test the pattern recognition code on the remaining images and display the recognised pattern on top of those images.

Is there a simple way of doing this?

Any help would be greatly appreciated.

Answer 1

[Walter Roberson]

No, there is no simple way of doing that.

[Pranay Koppula]

What would you recommend doing? How could I approach this problem? SImple or otherwise

[Walter Roberson]

If you threshhold the images, leaving the white centers as selected, then regionprops() to get the centroids. Use the x positions to crop the images to separate out the columns of water from each other.

But after that... it depends upon what you are trying to do. For example is the goal to examine photographs of houses in winter to detect icycles hanging down that might indicate heat loss ? Is the goal to examine photographs of stalegtites in caves to deduce the most prominent minerals ?