To realize a system for textile design patterns retrieval, we adopt an Image Indexing Method-Based Region. This indexing method is achieved by regions segmentation process followed by regions indexing one. Note that the later indexing process strongly depends on the quality of produced segmentation, and will have a negative impact on retrieval results. Therefore, an efficient segmentation method must be developed. In this paper, we propose an adaptive and efficient unsupervised color image segmentation method. The proposed method is based on Gaussian Mixture Model, and cluster validity index. The model parameters are estimated using EM algorithm, and the optimal cluster number is detected by Cluster Validity Index. Using this method, the design patterns are well extracted. In the second part of this work, the outcome of this segmentation results will be supplied to the input of the indexing process to complete our retrieval system. © The Textile Institute 2006. ________________________________________
A system for textile design patterns retrieval. Part I: Design patterns extraction by adaptive and efficient color image segmentation method
L D'Orazio;
2006
Abstract
To realize a system for textile design patterns retrieval, we adopt an Image Indexing Method-Based Region. This indexing method is achieved by regions segmentation process followed by regions indexing one. Note that the later indexing process strongly depends on the quality of produced segmentation, and will have a negative impact on retrieval results. Therefore, an efficient segmentation method must be developed. In this paper, we propose an adaptive and efficient unsupervised color image segmentation method. The proposed method is based on Gaussian Mixture Model, and cluster validity index. The model parameters are estimated using EM algorithm, and the optimal cluster number is detected by Cluster Validity Index. Using this method, the design patterns are well extracted. In the second part of this work, the outcome of this segmentation results will be supplied to the input of the indexing process to complete our retrieval system. © The Textile Institute 2006. ________________________________________I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.