Depth-based color stereo images retrieval using joint multivariate statistical models

摘要: Abstract The growing interest in using the three dimensional information various application fields has led to generation of huge color stereo image databases. As a result, it becomes necessary design efficient content-based retrieval systems well adapted indexing such large To this end, we propose paper different statistical-based approaches where associated estimated model parameters are considered as feature vector process. More precisely, Gaussian copula based multivariate Generalized will be used capture correlations existing images. While first strategy aims at exploiting cross-view cross-color channel redundancies, second one resorts more general joint statistical correlation between texture and depth information. Experimental results, performed on datasets, confirm benefits that can drawn from proposed approaches.