查看文章

Content-based image retrieval (CBIR) is a trivial task for multiple image processing systems, and has been a topic of research for over 2 decades. A wide variety of system models are proposed for this task, which range from simple feature correlation maximization to reinforcement deep learning via augmented models. But these models are based on context-specific CBIR applications, which limits their deployment capabilities for large-scale systems. In order to improve this scalability, a novel augmented feature extraction model Selective sampling with variance maximization & high-efficiency ensemble ranking (SSVMEF) is proposed in this text. The proposed model initially uses a large dataset for training, via extraction of color, edge, texture, Gabor, and deep learning features. These features have inherent redundancies, which are reduced via use of a customized variance function. This function estimates inter …