Unsupervised Remote Sensing Image Retrieval Using Probabilistic Latent Semantic Hashing

Unsupervised hashing methods have attracted considerable attention in large-scale remote sensing (RS) image retrieval, due to their capability for massive data processing with significantly reduced storage and computation. Although existing unsupervised are suitable operational applications, they exhibit limitations when accurately modeling the complex semantic content present RS images using binary codes (in an manner). To address this problem, letter, we introduce a novel method that takes advantage of generative nature probabilistic topic models encapsulate hidden patterns into final representation. Specifically, new latent (pLSH) model effectively learn hash three main steps: 1) grouping, where input archive is clustered several groups; 2) computation, pLSH used uncover highly descriptive from each group; 3) code generation, probability distributions thresholded generate codes. Our experimental results, obtained on two benchmark archives, reveal proposed outperforms state-of-the-art methods.