Robust Visual Place Recognition Based on Context Information

摘要

In large-scale and long-term visual SLAM, robust place recognition is essential for building aglobal consistent map. However，sensor viewpoints and environmental condition changes，includinglighting,weather，and seasons，bring a huge challenge to place recognition. We propose a placerecognition algorithm based on CNN features and graph model. Firstly,CNN features of images areextracted though an AlexNet network with migration characteristics, and N-nearest neighbor imagedescriptors of the current image descriptor are found by approximate nearest neighbor searching. Then,according to the difference between descriptors, a weighted directed acyclic graph(weighted DAG)model which describes a cost of context matching between images is established. Finally, a candidatematching sequence with minimum cost on this model is achieved by using Dijkstra algorithm.Comparedwith SeqCNiNSLAM and Fast-SeqSLAM, the experimental results demonstrate higher recognitionaccuracy and robustness of our algorithm.