Abstract
In this paper, an accurate segmentation approach to high resolution images based on wavelet-domain Gaussian Markov Random Field(GMRF) Tree Models is proposed. The novel wavelet decomposition algorithm and multi-scale segmentation of textured image are presented. This method captures the dependencies across the wavelet subbands and the interscale dependencies that are useful for texture analysis. The power of our technique lies in effective extraction of texture information in high resolution images. Experiments' prove the efficiency of the approach in the segmentation of high resolution images.
Original language | English |
---|---|
Pages | 721-724 |
Number of pages | 4 |
State | Published - 2004 |
Event | 2004 7th International Conference on Signal Processing Proceedings (ICSP'04) - Beijing, China Duration: 31 Aug 2004 → 4 Sep 2004 |
Conference
Conference | 2004 7th International Conference on Signal Processing Proceedings (ICSP'04) |
---|---|
Country/Territory | China |
City | Beijing |
Period | 31/08/04 → 4/09/04 |
Keywords
- GMRF tree models
- Segmentation
- Wavelet decomposition