Abstract

A creative re-ranking method named fusion similarity-based re-ranking is projected here to estimate the performance of synthetic aperture radar (SAR) image retrieval. First, the top ranked SAR images within the early retrieval results are selected for reranking. Considering the negative persuade of the speckle noise, three SAR-oriented visual features are chosen to represent them. In count, the diverse relevance scores consequent to an SAR image are analyzed in various modalities. Next, a fusion similarity is followed under the relevance score space to compute the equivalence between two SAR images. This fusion similarity is intended using the modal image matrix, which is constructed by the estimated scores to combine the contributions of all modalities. In conclusion, an existing re-ranking function is employed to re-rank the SAR images with the help of the expected scores and calculated fusion similarities. The positive investigational result shows that our re-ranking method is effective and efficient.

Paper Details