Scalable object instance recognition based on keygraph matching

• Published in: Pattern recognition letters Vol. 114; pp. 53 - 62
• Main Authors: Dazzi, Estephan, de Campos, Teofilo, Hilton, Adrian, Cesar, Roberto M.
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier B.V 15.10.2018; Elsevier Science Ltd
• Subjects: Affine transformations; Apexes; Clutter; Computer applications; Graph theory; Graph topological properties; Local invariant features; Mapping; Matching; Matching concept; Mathematical analysis; Object recognition; Occlusion; Pattern recognition; Semi-local graph matching; Topology; Local invariant features; 41A10; 65D05; 65D17; Graph topological properties; Semi-local graph matching; 41A05
• DOI: 10.1016/j.patrec.2017.10.038
• ISSN: 0167-8655

•A generalisation of local feature matching: keypoints are replaced by k-keygraphs.•Given keypoint matches, finding keygraph matches has a negligible computational cost.•With 4-keygraphs, more than 99% of the incorrect keypoint matches are eliminated.•4-keygraphs require 1% of the RANSAC iterations in comparison with 2-keygraphs. [Display omitted] We propose a generalisation of the local feature matching framework, where keypoints are replaced by k-keygraphs, i.e., isomorphic directed attributed graphs of cardinality k whose vertices are keypoints. Keygraphs have structural and topological properties which are discriminative and efficient to compute, based on graph edge length and orientation as well as vertex scale and orientation. Keypoint matching is performed based on descriptor similarity. Next, 2-keygraphs are calculated; as a result, the number of incorrect keypoint matches reduced in 75% (while the correct keypoint matches were preserved). Then, 3-keygraphs are calculated, followed by 4-keygraphs; this yielded a significant reduction of 99% in the number of remaining incorrect keypoint matches. The stage that finds 2-keygraphs has a computational cost equal to a small fraction of the cost of the keypoint matching stage, while the stages that find 3-keygraphs or 4-keygraphs have a negligible cost. In the final stage, RANSAC finds object poses represented as affine transformations mapping images. Our experiments concern large-scale object instance recognition subject to occlusion, background clutter and appearance changes. By using 4-keygraphs, RANSAC needed 1% of the iterations in comparison with 2-keygraphs or simple keypoints. As a result, using 4-keygraphs provided a better efficiency as well as allowed a larger number of initial keypoints matches to be established, which increased performance.