A Fast Incremental Spectral Clustering Algorithm for Image Segmentation
| Autor: | Chenyu Chang, Xia Li Wang, Xiaochun Wang |
|---|---|
| Rok vydání: | 2017 |
| Předmět: |
business.industry
Computer science Pattern recognition 02 engineering and technology Image segmentation Spectral clustering ComputingMethodologies_PATTERNRECOGNITION 020204 information systems Linear algebra 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Artificial intelligence business Cluster analysis |
| Zdroj: | 2017 International Conference on Computational Science and Computational Intelligence (CSCI). |
| DOI: | 10.1109/csci.2017.68 |
| Popis: | Clustering aims at grouping a given set of data points into a number of clusters without resorting to any a priori knowledge. Due to its important applications in data mining, many techniques have been developed for clustering. Being one of the most popular modern clustering algorithms, spectral clustering is simple to implement, can be solved efficiently by standard linear algebra software, and very often outperforms traditional clustering algorithms such as the k-means algorithm. However, it is not very well scalable to modern large datasets which typically have millions of items. To partially circumvent this drawback, in this paper, we propose an integration-based fast incremental spectral clustering algorithm which is particularly designed for image segmentation tasks. The algorithm first divides a given large dataset into several smaller partitions, next applies spectrum clustering to each partition, and finally integrates them using a BIRCH tree. Experiments performed on image data demonstrate the efficacy of our method. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|