Following the advertizing statements of several manufacturers there are several benefits using square-shaped phosphor-screens for EBSD pattern detection. Any truth to the rumour? Or is the used shape irrelevant since the commonly used Hough (or Radon) transform is anyway based on a circular area which needs to be defined?
Due to the fact that the patterns are radon- oor hough-transformed only a circular pattern part can be transformed. But for square type screens you can select your transformed part from a larger screen area so perhaps you can optimze the transformed are to reduce artefacts from projection geometry so that you principly have an larger area to chose from. But if you can detect also from the whole screen very good patterns and peaks in the tranformed space i think the angular resolution of your EBSD should improve like when you "zoom out" by reducing the detector sample distance.
I think I don't understand your answer correctly: You say that only circular patterns can be transformed. But in the next sentences you offer an opportunity that transformed parts which are actually outside the circular area but can be used to improve the information, e.g. angular resolution?
The input of the transformation is always a circular area. If the screen is circular the whole screen can be taken into account. Using a square screen offers the opportunity to also transform parts that are out of the middle circle if the circular mask exceeds the screen but some artefacts arise to the Hough space then. Fortunately these artefacts were not detected by the peak identification so that with a square sreen you are more flexible to chose different areas of the patterns or larger areas without reducing the detector distance.
Well, I am not absolutely but quite sure that everything what contributes as peak in Hough transform (which are not artifacts) will be used in band detection and indexing finally. The attached files display both the transform of the circular area as well as the complete transform of the rectangular shaped EBSD patterns. I hope you can agree that the additional peaks do not look like artifacts and can be very beneficial for indexing of low-symmetric phases.
Yes that is. In the third image you transformed the whole screen and theblack areas are the artefacts that are not detected by a the peak detection system. The addtional peaks are a result from the llines that are longer at the square screen than with a circular mask. Have both Hough spaces really the same resolution? The peak height naturally also depends on teh Hough resolution.So the main advantage: You can transform the whole screen - > small lines with a circular mask are longer in the square case -> additional peaks you can index. But do you really use more than 11 peaks that are easily visible at the upper Hough space?
I see, the black areas are the artifacts you want to point out. My misunderstanding. I call it simply zero-area where no image exists which cannot be transformed.
Coming back to your other question: no, both Hough transforms don't have the same resolution. I only wanted use them to show the difference between full and partial Hough transform.
And yes, we are using more than 11 bands. However, there are infinite Hough peaks but only some are well visible because of their hight. In the third image you have at least three very strong peaks which do not appear in the second image. The phase presented is cubic, If you have low-symmetric phases the best visible bands might be outside of your circular area which causes misindexings or even no indexings. This also happens if high-symmetric zone axis are in the center of your circular area and there are no other strong bands close to the zone axis. Well, you can shift your sample closer to the screen what already has been mentioned in one of the former comments.
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Materials Science