I have research on segment 2D teeth plaster model. We have tried using semi supervised method (GrabCut), and the results are satisfying. However, we would like to make the segmentation process fully automatic. We have made experiment using clustering method, edge detection method, and adaptive thresholding but the experiment result is not satisfying. What is the right method for segmenting the teeth image?
Here is a sample of the dental 2D plaster model.
Hi Dear
I implement procedure on your images , the results is attached with answer
Procedure is : Sharpening + 2 pass Gray Dilation Morphology + 1 pass Binary erosion Morphology
Thanks
Hi Arie,
You can try the active contour, else I would recommend using machine learning to segment the teeth. When I was in Vienna attending a summer school, they showed how random forests (after training) could robustly detect fingertips in a hand scan image.
Hi Arie,
I think contrast limited adaptive histogram equalization (CLAHE) might help you as preprocessing step, it is available as a built in function in MatLab (adapthisteq)
Hi Arie,
you can try using deep learning. or try Histogram of Oriented Gradient (HOG)
- Badges
- Science topic