Hierarchical Multi-Process Fusion for Visual Place Recognition

Abstract

A recent approach to the Visual Place Recognition (VPR) problem has been to fuse the place recognition estimates of multiple complementary VPR techniques simultaneously. However, selecting the optimal set of techniques to use in a specific deployment environment a-priori is a difficult and unresolved challenge. Further, to the best of our knowledge, no method exists which can select a set of techniques on a frame-by-frame basis in response to image-to-image variations. In this work, we propose an unsupervised algorithm that finds the most robust set of VPR techniques to use in the current deployment environment, on a frame-by-frame basis. The selection of techniques is determined by an analysis of the similarity scores between the current query image and the collection of database images and does not require ground-truth information. We demonstrate our approach on a wide variety of datasets and VPR techniques and show that the proposed dynamic multi-process fusion (Dyn-MPF) has superior VPR performance compared to a variety of challenging competitive methods, some of which are given an unfair advantage through access to the ground-truth information.

Improving Visual Place Recognition Performance by Maximising Complementarity

“…Similarly, a multi-process fusion system is introduced in [19] which combines multiple VPR methods using a Hidden Markov Model (HMM) to identify the optimal estimated location over a sequence of images. The authors of [20] have presented a three-tier hierarchical multi-process fusion system which is customizable and may be extended to any arbitrary number of tiers. A different place recognition method is used in each tier to compare the query image with the provided sequence of images.…”

“…Recently, a new approach named multi-process fusion has been introduced that combines several image processing methods and negates the requirement of multiple sensors to improve VPR performance [19], [20]. The concept comes from the empirical data which suggests that some VPR methods are more suitable for certain types of environments and scenarios than others [10].…”

“…Hence, utilising multiple VPR techniques simultaneously may compensate for each other's weaknesses. Although the systems presented in [19], [20] exhibit promising results, they do not provide a well-defined criterion for selection of VPR techniques based on complementarity out of the available options. Supposing that the fused VPR methods will complement each other in all cases is not a valid assumption and may have detrimental effects on performance and computation.…”

Explicit feature disentanglement for visual place recognition across appearance changes

“…Typical pipeline for visual place recognition includes (1) feature extraction and (2) matching, 5,6,[16][17][18][19][20][21] optionally followed by (3) temporal fusion. 7,8,22,23 This article will focus on the feature extraction module.…”

Hierarchical Visual Place Recognition Based on Semantic-Aggregation

“…Emilio et al [33] proposed an appearance-based method for topological mapping based on hierarchical decomposition of environment, and proved that the hierarchical method could reduce search space in identifying place while improve mapping accuracy in creating a map. Stephen and Milford [34] developed a new stacked hierarchical localization framework, which concatenated localization hypotheses from techniques with complementary characteristics at each layer, performing well on two challenging datasets. These works have proved that hierarchical strategy is useful and effective in reducing search space and localization.…”