I appreciate any clarifications regarding this(python source code, articles "although I read many",...):
The main thing I want to make is mesh->graph->weight function -> descriptors -> featured connected components -> matching
I'm trying to obtain a new shape descriptor dependable on simple descriptors like "shape index, mean curvature".
First I have obtained from my mesh Interest points using 3D Harris detector then obtain descriptor for these interest points like this example: Mesh1, Mesh2(is rotated, or scaled, or transformed Mesh1) I have for each ip1,ip2 (interest points) and for each "ip" I obtained the descriptors so if I have for example 400 interest points, I have a vector with 400 values, that represent the descriptors for these points: [2 4 15 18 90 ...] these are "ip1" and d1 = [0.3 0.002 0.0 1 0.543 ....] where 0.3 is, for example, the mean curvature of interest point (2) and so on, and what I tend now to do is to evaluate the descriptors, I read about using Euclidean distance for the vectors, but not sure about it.
I already asked such a question (https://scicomp.stackexchange.com/questions/34376/shrec-2010-descriptors)
I'm coding using python, and I have mesh, not images.
I suggest you a helpful link for your query's solution as: https://www.analyticsvidhya.com/blog/2019/08/3-techniques-extract-features-from-image-data-machine-learning-python/
Dear Sir Muhammad Ali ,
Sorry but I mentioned before my work is on 3D objects(mesh and points clouds), and really I'm not able to relate between the two, images and meshes. As well in the link attached, it's all about extracting the feature, in my work, I did extract features as feature vector as mentioned [0.3 0.002 0.0 1 0.543 ....] for the feature points in the mesh already obtained as well, but what I'm seeking for, having two mesh and needs to match them (or find correspondence ) between the feature vectors.
Thank you
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Programming Languages
- Python