A Context-Aware Locality Measure for Inlier Pool Enrichment in Stepwise Image Registration

Zhang, Su; Zhao, Wanjing; Hao, Xuying; Yang, Yang; Guan, Cuntai

doi:10.1109/tip.2019.2961480

Cited by 19 publications

(4 citation statements)

References 53 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the third series of experiments, we conducted quantitative comparison with six state-of-the-art algorithms: GLPM [40], LPM [39], ICF [20], GS [58], LMR [52], and SIR [41]. Then, 27 pairs of yaw, pitch, and mixture scenarios in the dataset are used for testing.…”

Section: Results Of Comprehensive Quantitative Comparisonmentioning

confidence: 99%

“…In the fourth series of experiments, considering the diversity of UAV images, we add an extreme scenario, in which not only rotation occurred, but also low overlap, distortion, and scaling are combined, thereby increasing the difficulty of registration. We conducted quantitative comparison of extreme scenarios with six state-of-the-art algorithms: GLPM [40], LPM [39], ICF [20], GS [58], LMR [52], SIR [41]. Then, 23 pairs of images on extreme scenarios were used for testing.…”

Section: Results Of Quantitative Comparison Of Extreme Scenariosmentioning

confidence: 99%

“…Its variant-guided locality preservation matching (GLPM) [40] further employs a small putative set with a high inlier ratio to guide the matching. SIR [41] enriches the inlier pool and refines the transformation by repeated iteration to select the inliers from the candidate pool, and the refined transformation prunes inconsistent mismatches to alleviate the incoming matching ambiguity. RFM-SCAN [42] can adaptively cluster a set of putative matches into several inlier groups with motion consistency and a outlier group with linearithmic time complexity.…”

Section: Related Workmentioning

confidence: 99%

“…Gaussian high-pass filter in the algorithm. In the second category, we evaluate the performance of our algorithm against six state-of-the-art methods: GLPM [40], LPM [39], ICF [20], GS [58], LMR [52], and SIR [41] in the comprehensive rotation and extreme cases for feature matching and registration. The experiments are conducted on a desktop with 3.0 GHz Intel Core i5-8500 CPU, 16-GB RAM, and MATLAB code.…”

Section: Implementation Detailsmentioning

confidence: 99%

See 3 more Smart Citations

Rotation-Invariant Siamese Network for Low-Altitude Remote-Sensing Image Registration

Liu

Gong

Chen

et al. 2020

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

Self Cite

View full text Add to dashboard Cite

Multiple-view change caused by small unmanned aerial vehicles (UAVs) monitoring the ground, result in image distortion, multi-view transformation and low overlap. Thus, such change has a strong effect on the accuracy of image registration. In this study, we utilize a Siamese network to deal with the complexity registration of low-altitude remote-sensing images. A robust neighbor-guided patch representation is designed to describe feature points based on neighborhood relation reconstruction and patch selection. The network is trained based on rotation-invariant layer to solve the inevitable rotation and nonrigid deformation caused by multi-view images in low-altitude remote-sensing images. With only 3 training images involving 4,500 putative matches, the experiment results demonstrated that the learned network can process the scenarios of yaw rotation, pitch rotation, mixture, and extreme(e.g. mixture, scaling and distortion occur simultaneously) of UAV better than other six state-of-the-art methods.

show abstract