What is the appropriate Neuarl Network for classifying AFM force curves?

1 次查看（过去 30 天）

Michael Keeling 2019-10-26

1
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/487549-what-is-the-appropriate-neuarl-network-for-classifying-afm-force-curves

评论： Michael Keeling 2019-10-28

My data consists of force curves taken in a grid over various materials, of which I know the location. I would like to classify the force curves as belonging to a material. I'd assume I'd need to use something like patternnet for that. However the spatial distribution of the curves also matters, i.e. 2 curves next to each other are likely from the same material. What network could I use for this?

Thank you!

2 个评论
显示无隐藏无

Image Analyst 2019-10-26

Can you show plots of some of the curves? And I don't understand that two curves being "next to each other" means they are likely from the same material. What exactly does that mean?

Michael Keeling 2019-10-28

Please find an example curve below (normalised by zscore).

Regarding the materials: there is an island of material A inside material B. Since the force curves are taken in a grid pattern they basically form an image where each pixel is a curve. Therefore a force curve obtained in the middle of the island of Material A would only be surrounded by force curves that are also of Material A.

Basically I want to perform image segmentation, but my images have pixels that are actually force curves.