- Tytuł:
- Wound image segmentation using clustering based algorithms
- Autorzy:
-
Farmaha, Ihor
Banaś, Marian
Savchyn, Vasyl
Lukashchuk, Bohdan
Farmaha, Taras - Powiązania:
- https://bibliotekanauki.pl/articles/2064381.pdf
- Data publikacji:
- 2019
- Wydawca:
- STE GROUP
- Tematy:
-
clustering
Segmentation
machine learning
neural networks
wounds
segmentacja
nauczanie maszynowe
sieci neuronowe
rany
klastrowanie - Opis:
- Classic methods of measurement and analysis of the wounds on the images are very time consuming and inaccurate. Automation of this process will improve measurement accuracy and speed up the process. Research is aimed to create an algorithm based on machine learning for automated segmentation based on clustering algorithms Methods. Algorithms used: SLIC (Simple Linear Iterative Clustering), Deep Embedded Clustering (that is based on artificial neural networks and k-means). Because of insufficient amount of labeled data, classification with artificial neural networks can`t reach good results. Clustering, on the other hand is an unsupervised learning technique and doesn`t need human interaction. Combination of traditional clustering methods for image segmentation with artificial neural networks leads to combination of advantages of both of them. Preliminary step to adapt Deep Embedded Clustering to work with bio-medical images is introduced and is based on SLIC algorithm for image segmentation. Segmentation with this method, after model training, leads to better results than with traditional SLIC.
- Źródło:
-
New Trends in Production Engineering; 2019, 2, 1; 570--578
2545-2843 - Pojawia się w:
- New Trends in Production Engineering
- Dostawca treści:
- Biblioteka Nauki
Artykuł