A Novel Image Retrieval Method for Image Based Localization in Large-Scale Environment
Xiliang Yin, Lin Ma, Xuezhi Tan · 2021
Recently, indoor localization has become a hot research field. Although there are already different localization technologies, image-based localization is attracting more attention from researchers due to its unique advantages. The image retrieval algorithm is still a key part of the existing image-based localization system in a large scale environment. It is a challenging task for image retrieval to achieve a balance among image retrieval ratio, retrieval accuracy, and time complexity under such an environment. The state of the art image retrieval methods for image-based localization system still suffers from these problems. Therefore, we propose a novel image retrieval method aiming at reducing the time complexity without losing the retrieval accuracy. We compare the performance of the proposed method with the popular algorithms including the state of art in the indoor localization field, moreover, they are validated in a public dataset. It demonstrates that our method provides better results, which is verified by the well-known image recognition and visual localization algorithm.