PROJECT TITLE :

A three-layered graph-based learning approach For remote sensing image retrieval - 2016

ABSTRACT:

With the emergence of big volumes of high-resolution remote sensing pictures created by all sorts of satellites and airborne sensors, processing and analysis of these images need effective retrieval techniques. To alleviate the dramatic variation of the retrieval accuracy among queries caused by the one image feature algorithms, we have a tendency to developed a completely unique graph-based mostly learning technique for effectively retrieving remote sensing images. The tactic utilizes a three-layer framework that integrates the strengths of question growth and fusion of holistic and local options. In the first layer, two retrieval image sets are obtained by, respectively, using the retrieval strategies based mostly on holistic and native features, and therefore the top-ranked and common pictures from both of the prime candidate lists subsequently type graph anchors. In the second layer, the graph anchors as an growth query retrieve six image sets from the image database using each individual feature. Within the third layer, the photographs within the six image sets are evaluated for generating positive and negative knowledge, and SimpleMKL is applied to be told appropriate question-dependent fusion weights for achieving the final image retrieval result. Intensive experiments were performed on the UC Merced Land Use-Land Cover information set. The supply code has been accessible at our web site. Compared with alternative related methods, the retrieval precision is considerably enhanced without sacrificing the scalability of our approach.