Efficient Representation of Local Geometry for Large Scale Object Retrieval

State of the art methods for image and object re- trieval exploit both appearance (via visual words) and local geometry (spatial extent, relative pose). In large scale problems, memory becomes a limiting factor { lo- cal geometry is stored for each feature detected in each image and requires storage larger than the inverted le and term frequency and inverted document frequency weights together. We propose a novel method for learning discretized local geometry representation based on minimization of average reprojection error in the space of ellipses. The representation requires only 24 bits per feature without drop in performance. Additionally, we show that if the gravity vector assumption is used consistently from the feature description to spatial verication, it improves retrieval performance and decreases the memory foot- print. The proposed method outperforms state of the art retrieval algorithms in a standard image retrieval benchmark.

• Type: Conference
• Year: 2009
• Where: CVPR
• Authors: Michal Perdoch (Czech Technical University), Ondrej Chum (Czech Technical University), Jiri Matas (Czech Technical University)
• Keywords: Computer Vision, CVPR 2009, Large Scale Object Retrieval