Advanced Search

2000x Microscope

Viewing 12 posts - 31 through 42 (of 42 total)
  • Author
    Posts
  • #5855
    Anthony Yan
    Participant
    • Topics: 4
    • Replies: 96

    Hi Clay,

    Wow, that last image is amazing! More please. 🙂

    btw, I think the forums are re-sizing your images, so the measurements/annotations you made to the image are too small to really see or read. Is there a way for you to show us those images without so much resizing? Maybe post them to your blog and then link that?

    You also mentioned that CoorsTek gives measurements for surface-roughness of their ceramics in Ra. I’m not sure how they do those measurements, but I’m feeling unsure that they would use a contact profilometer to do it. This is because the extremely hard abrasives might wear out the stylus. The stylus that is dragged over the surface needs to be very sharp and precise. If it were dragged out along any very abrasive material, I think it would be damaged, or at least worn out. Maybe it’s a diamond stylus. I don’t know.

    My guess (just a guess!) is that they use some type of non-contact profilometer to do their measurements. Since they are a huge company, one of the largest producers of technical ceramics, I would suppose they have invested in high-end optical (non-contact) profilometers and other fancy machines for doing surface measurements. Don’t know if they would tell you, but maybe you could ask them out of curiosity? 🙂

    If you don’t mind me getting a bit philosophical:

    I also only recently learned about profilometers. I call this an example of not-knowing what I don’t know. For example, I don’t know how to speak German, but I know enough to realize that, and I also know where I could go to learn to speak German. This is an example of knowing what I don’t know. I think this is an important point for two reasons:

    If I know what I don’t know, then:
    (1) I know how to gain more expertise.
    (2) I can begin to assess my own level of expertise.

    It’s very difficult to do (1) and (2) if you don’t know what you don’t know. There is always trial-and-error… But it’s probably better to leverage huge areas of research which have already been done.

    One way I’m trying to figure out what I don’t know, is to read up on engineering. Engineers and manufacturers have had to measure cutting performance, material strength, friction, and surface properties for centuries (even if you start counting as late as the Industrial Revolution, not to mention the ancient Egyptians). So along these lines, I’ve been browsing engineering, manufacturing, and science topics in places like Wikipedia, research papers, the library, YouTube, Amazon.com, and company websites. Ideally, it would be great to also do some sharpening tests, and talk to knife makers, etc.

    I would humbly invite others to do the same, and then share with us what they find.

    Sincerely,
    –Lagrangian

    #5857
    wickededge
    Keymaster
    • Topics: 123
    • Replies: 2936

    Lagrangian,

    As always, I love reading your posts and hearing your perspective. I think I can ask CoorsTek directly about what they’re using. As for me, I’m grateful to everyone that occasionally hits me over the head and says “hey dummy, somebody already thought of that and there’s a well defined method in place!” I’ll continue to keep looking and hopefully will adopt some of the wisdom of your approach and I will, of course share.

    My blog also shrinks the images but maybe I can stash them on Dropbox or a similar service and link the posts to the full sized images.

    –Clay

    -Clay

    #5868
    Phil Pasteur
    Participant
    • Topics: 10
    • Replies: 943

    Hi Clay,

    Wow, that last image is amazing! More please. 🙂

    btw, I think the forums are re-sizing your images, so the measurements/annotations you made to the image are too small to really see or read. Is there a way for you to show us those images without so much resizing? Maybe post them to your blog and then link that?

    Sincerely,
    –Lagrangian

    Hey Anthony,
    Just a data point, I am seeing the notation perfectly here. I am linking a small capture of the corner of one of Clay’s images. If this is what you can’t see…

    I wonder if it is your browser or a setting that you are using?

    Phil

    Attachments:
    #5869
    Tom Whittington
    Participant
    • Topics: 4
    • Replies: 159

    If anyone is having difficulty viewing the images, maybe due to scaling being weird or cutting off the image, you can try grabbing the raw image by clicking and dragging it onto the address bar/tab list. Forum software likes to scale stuff pretty aggressively to save server resources, so even the raw image posted through the forum itself will be scaled down to whatever the built in size limit is.

    They look fine on my end using Chrome and a 1920×1200 display, just as another reference point.

    #5871
    Anthony Yan
    Participant
    • Topics: 4
    • Replies: 96

    In one of Clay’s images, he has very very tiny annotations. I can barely see them, and they are unreadable to me. I took Clay’s image and marked where I see the annotations in red.

    Clay’s original image is here.

    This is the same image with the areas of interest marked in red:

    I was hoping to read what these were. Are you guys able to read these? I’m using Internet Explorer 9 in Windows 7.

    Sincerely,
    –Lagrangian

    #5872
    Phil Pasteur
    Participant
    • Topics: 10
    • Replies: 943

    Heck no I can’t read them…
    I did not even see them before you circled them in red…
    Would have never known that they were notations…
    Phil

    #5873
    Tom Whittington
    Participant
    • Topics: 4
    • Replies: 159

    I didn’t notice those either, good catch! With the full resolution images they’ll be legible I’m sure. Hopefully Clay will work out a quick and easy hosting solution!

    #5885
    wickededge
    Keymaster
    • Topics: 123
    • Replies: 2936

    I added the image to Dropbox: Superfine 2000x

    -Clay

    #5891
    Anthony Yan
    Participant
    • Topics: 4
    • Replies: 96

    Thanks Clay! It is super interesting to see those detailed measurements. 🙂

    #5892
    wickededge
    Keymaster
    • Topics: 123
    • Replies: 2936

    It’s challenging to pick out starting and ending points for the measurement so suggestions are most welcome. It is some consolation to know that the grooves do match the stated grits of the stones and strops so far. Hopefully it’s not a case of finding what you’re already looking for…

    Thanks Clay! It is super interesting to see those detailed measurements. 🙂

    -Clay

    #5895
    Josh
    Participant
    • Topics: 89
    • Replies: 1671

    Sweet! I love this thread! Our world just opened up gentlemen! Clay, I saw in another post by Staze that you posted the below pics… are they looking down directly at the edge from above?! Can’t wait to have you play around w/ this… your photo skills are improving day by day 😀


    #5898
    Anthony Yan
    Participant
    • Topics: 4
    • Replies: 96

    Hi Clay,

    After seeing your recent images, I’ve been wondering about the same thing. Probably we can apply some methods from Computer Vision to measure scratch widths. For example, there are open-source software libraries like OpenCV for doing computer vision.

    I can’t promise much, because I haven’t much free time, and it’s often not clear if the algorithms will give a clear and definitive measurement. The advantage is that an algorithm will give an unbiased answer. More precisely, if the algorithm is biased (such as a tendency to underestimate measurements), at least that bias will tend to be consistent and reproducible.

    I’m reminded of the tennis player Andre Agassi. He complained that even the latest high-tech ball-tracking computers would make incorrect line calls. But he was consoled by the fact that the calls were made by a machine that has no bias as to who should win the match.

    Sincerely,
    –Lagrangian

    P.S. For those with a technical background:
    To analyze the images for scratch width, I have some general ideas right now:
    (1) There are many papers about computer image-analysis in the medical industry. For example, given a tissue sample, these algorithms estimate things like the distribution of diameters of blood-vessles, etc.
    (2) For scratch patterns which are parallel, maybe we could simply apply Fourier Analysis and look at the distribution of harmonic components in the frequency domain. It probably won’t happen, but if we are _super lucky_ there would be a dominant peak at the average scratch size.
    (3) There are various edge-detection methods in computer vision, such as the Canny edge detector, and things in computer vision called “snakes”. A snake is basically a line or curve that is iteratively adjusted to find an edge or feature in an image. So it “wiggles” into position.
    (4) There are a bazillion variants of these ideas, but we should probably just try the simplest ones (unless some of them are already implimented in free public packages, such as OpenCV).
    (5) It can be tricky to estimate the accuracy and precision (ie: uncertainty) of the results, especially when we do not have test cases where we know ground-truth.
    (6) Computer vision is much more difficult than would naively appear; many algorithms are extremely sensitive to signal-to-noise ratios, small changes in lighting direction, focus, and so on. Consider that in the 1960’s, people thought computer facial recognition would be solved in a couple of years. It is only 40+ years later, that computer facial recognition is mature enough to be put in consumer cameras, and used in some security systems. And even then, facial recognition is often not reliable.
    (7) Computer-vision might be the wrong tool for this problem. Instead, perhaps, measurements from a contact profilometer might be the most direct answer. If the contact-profilometer can show us the trace of a single scan-line, that would tell us a lot! Engineers in Surface Metrology must have good ways of doing such measurements.

Viewing 12 posts - 31 through 42 (of 42 total)
  • You must be logged in to reply to this topic.