Sparse3D: A new global model for matching sparse RGB-D dataset with small inter-frame overlap

Le, Canyu; Li, Xin

doi:10.1016/j.cad.2018.04.018

Cited by 10 publications

(11 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A widely adopted consensus constraint is loop closures: correct pairwise alignments support each other spatially and their relative transformations compose to identity along a closed loop if they are considered in a loop. Such a loop closure constraints have been widely applied on refining or pruning pairwise alignments in many vision-based SLAM and environment reconstruction like [11], [12], [25]. However, enforcing loop closures in jigsaw puzzle solving problem is more challenging than it is in these SLAM and reconstruction tasks.…”

Section: Overviewmentioning

confidence: 99%

“…Based on the above definitions, we can formulate the global composition as an optimization problem in Eq. (12).…”

Section: A Terminologies and Formulationsmentioning

confidence: 99%

“…The loop closure constraint is implicitly included in Eq. (12). Since the edges/alignments are consistent in a closed loop, the optimization of Eq.…”

Section: A Terminologies and Formulationsmentioning

confidence: 99%

“…3) Accuracy and Complexity Analysis on HLM: The HLM algorithm aims to extract as many compatible loops as possible from the given multi-graph G. This is consistent with minimizing Eq. (12). Maximizing the number of compatible loops selected will minimize the second (indicator variable penalty) term, and since these loops are compatible to each other, it does not increase the first term.…”

Section: A Hierarchical Loop Merging (Hlm) Algorithmmentioning

confidence: 99%

“…CNN+BF INDICATES THE STRATEGY OF USING PROPOSED CNN COMPATIBILITY DETECTOR IN ALIGNMENT PRUNING, FOLLOWED BY A BEST FIRST SEARCH IN GLOBAL COMPOSITION. GLC AND HLM STANDS FOR THE GREEDY LOOP CLOSING[12] AND HIERARCHICAL LOOP MERGING STRATEGIES, RESPECTIVELY. Compared with the greedy strategy, the GLC and HLM strategies produce more reliable results because of the usage of loop closure constraints.GLC versus HLM.…”

mentioning

confidence: 99%

See 4 more Smart Citations

JigsawNet: Shredded Image Reassembly Using Convolutional Neural Network and Loop-Based Composition

2019

IEEE Trans. on Image Process.

Self Cite

View full text Add to dashboard Cite

This paper proposes a novel algorithm to reassemble an arbitrarily shredded image to its original status. Existing reassembly pipelines commonly consist of a local matching stage and a global compositions stage. In the local stage, a key challenge in fragment reassembly is to reliably compute and identify correct pairwise matching, for which most existing algorithms use handcrafted features, and hence, cannot reliably handle complicated puzzles. We build a deep convolutional neural network to detect the compatibility of a pairwise stitching, and use it to prune computed pairwise matches. To improve the network efficiency and accuracy, we transfer the calculation of CNN to the stitching region and apply a boost training strategy. In the global composition stage, we modify the commonly adopted greedy edge selection strategies to two new loop closure based searching algorithms. Extensive experiments show that our algorithm significantly outperforms existing methods on solving various puzzles, especially those challenging ones with many fragment pieces. Data and code have been made available in https://github.com/Lecanyu/JigsawNet.

show abstract

Section: Overviewmentioning

confidence: 99%

“…Based on the above definitions, we can formulate the global composition as an optimization problem in Eq. (12).…”

Section: A Terminologies and Formulationsmentioning

confidence: 99%

“…The loop closure constraint is implicitly included in Eq. (12). Since the edges/alignments are consistent in a closed loop, the optimization of Eq.…”

Section: A Terminologies and Formulationsmentioning

confidence: 99%

Section: A Hierarchical Loop Merging (Hlm) Algorithmmentioning

confidence: 99%

mentioning

confidence: 99%

See 3 more Smart Citations