2/27/24. Steel Weld Pictures

Frank L -

As I previously thought, the program that I created for turning the steel weld data into pictures wasn’t really that complicated. The actual program itself was only about 20 lines of code, so super basic and simple. I spent most of the time just learning the principles behind what was happening and figuring out which functions to use. After that, it was a matter of properly applying them. One of the main resources I used was Matplotlib, a library of prewritten code in python that is designed for creating graphs, charts, and basically everything that has to do with visualizing images from data. I also had to import other libraries that had the functions for reading through and extracting numbers from excel files, but those weren’t too complicated either.

This is one of the pictures that I was able to create using the visualization program. It’s a 2D rendition of a steel to steel resistance spot weld. Just to understand what’s happening here, a resistance spot welding is a type of manufacturing process where two sheets of metal are welded together by applying heat and pressure with an electric current to certain points. The colors in the pictures are meant to signify the surface height of the weld. The darker and deeper colors indicate that the steel sheet is deeper at those certain points, while the lighter colors represent the opposite.

As for identifying cracks within the steel, I have no clue how that’s done. In fact, I was just informed that this entire subtask has been passed on to another student, so it’s out of my hands now. Yesterday, Hasnaa gave me a new task to help her out with a different project she’s working on, so I’ve moved on to learning how to make a program for cropping images based on contour detection. Looking forward to giving an update on that next post.

More Posts


All viewpoints are welcome but profane, threatening, disrespectful, or harassing comments will not be tolerated and are subject to moderation up to, and including, full deletion.

    Do your have any idea of what your final product is going to be, and how you are going to apply knowledge from a task such as this into your product?
    That very cool to see a visual of a spot weld, but sad to hear that you had the project taken out of your hands. As for your next project, what is contour detection?
    If the crack project was still in your hands do you know if it'd be similar to scratch detection algorithms in image manipulation programs like GIMP (especially since most libraries for it essentially just look for abnormal changes in greyscale values) or Photoshop?
    What program does it use?
    I'm thinking of making some sort of AI program as my final product, so I'm learning more about machine learning algorithms on my own and building off the things that I learn about in my role as a research assistant. I'll talk more in-depth about what the goal for my final product will be in a later post.
    Contour detection works by identifying edges where the pixels change in intensity. It sounds complicated but I'm basically just cropping images to get rid of unnecessary background space. We could crop the images manually, but with a computer program, it'd take a few seconds to perfectly crop hundreds of images.
    From what I understood, the steel weld images were just created as a 2D representation for people to look at. The actual crack identification AI program would read through the depth numbers in the excel files to find certain types of patterns. It might be based on the same types of principles that GIMP and Photoshop use though.
    I'll be creating my own python program for the contour detection task.

Leave a Reply

Your email address will not be published. Required fields are marked *