A New Image Decomposition Method for Content-Based Remote Sensing Image Retrieval

Firstly,the ordinary image decomposition method are listed: Quad-tree,Quin-tree and Nona-tree,then these methods are analyzed and compared,and the concepts of average block cover ratio and redundant data ratio are proposed.On the basis of these,a judgment is made that the average block ratio and redundant data ratio are the most important standards to estimate the decomposition quality,the improve room of Quin-tree is discovered.Then,a new block-method,Quin~(+)-tree is proposed.Lastly,a remote sensing image A,1 024 pixels width by 1 024 pixels height is chosem and decompose by Quad-tree,Quin-tree,Quin~(+)-tree and Nona-tree by three levels.From five different aspects,these four image decomposition methods' results are analyzed and compared.