Outdoors augmented reality on mobile phone using loxel-based visual feature organization 论文

摘要

Abstract—We have built an outdoors augmented reality system for mobile phones that matches camera-phone images against a large database of location-tagged images using a robust image retrieval algorithm. We avoid network latency by implementing the algorithm on the phone and deliver excellent performance by adapting a state-of-the-art image retrieval algorithm based on robust local descriptors. Matching is performed against a database of highly relevant features, which is continuously updated to reflect changes in the environment. We achieve fast updates and scalability by organizing image descriptors on a locationbased grid of loxels, which enables pruning of irrelevant features based on proximity to the user. It also permits incrementally updating the features stored on the phone, making the system amenable to low-bandwidth wireless connections. We further reduce the number of features through a clustering scheme that combines redundant features into meta-features and removes outliers. Additional memory and bandwidth savings come from improved coding efficiency of descriptors. We demonstrate system robustness on several datasets of location-tagged images and show a smart-phone implementation that achieves a high image matching rate while operating in near real-time. Index Terms—real-world image annotation, robust visual feature descriptors, mobile augmented reality, large-scale image retrieval systems I.