Quasi-Dense Matching for Oblique Stereo Images through Semantic Segmentation and Local Feature Enhancement
Open Access
- 8 February 2024
- journal article
- research article
- Published by MDPI AG in Remote Sensing
- Vol. 16 (4) , 632
Abstract
This paper proposes a quasi-dense feature matching algorithm that combines image semantic segmentation and local feature enhancement networks to address the problem of the poor matching of image features because of complex distortions, considerable occlusions, and a lack of texture on large oblique stereo images. First, a small amount of typical complex scene data are used to train the VGG16-UNet, followed by completing the semantic segmentation of multiplanar scenes across large oblique images. Subsequently, the prediction results of the segmentation are subjected to local adaptive optimization to obtain high-precision semantic segmentation results for each planar scene. Afterward, the LoFTR (Local Feature Matching with Transformers) strategy is used for scene matching, enabling enhanced matching for regions with poor local texture in the corresponding planes. The proposed method was tested on low-altitude large baseline stereo images of complex scenes and compared with five classical matching methods. Results reveal that the proposed method exhibits considerable advantages in terms of the number of correct matches, correct rate of matches, matching accuracy, and spatial distribution of corresponding points. Moreover, it is well-suitable for quasi-dense matching tasks of large baseline stereo images in complex scenes with considerable viewpoint variations.Keywords
Funding Information
- National Natural Science Foundation of China (42171435, ZR2021MD006, 2023M732686, Z2021014, SDYKC2022151)
- Shandong Provincial Natural Science Foundation (42171435, ZR2021MD006, 2023M732686, Z2021014, SDYKC2022151)
- China Postdoctoral Science Foundation (42171435, ZR2021MD006, 2023M732686, Z2021014, SDYKC2022151)
- Undergraduate Education and Teaching Reform Foundation of Shandong Province (42171435, ZR2021MD006, 2023M732686, Z2021014, SDYKC2022151)
- high quality graduate course of Shandong Province (42171435, ZR2021MD006, 2023M732686, Z2021014, SDYKC2022151)
This publication has 23 references indexed in Scilit:
- DCAN: Dual Channel-Wise Alignment Networks for Unsupervised Scene Adaptation2018
- L2-Net: Deep Learning of Discriminative Patch Descriptor in Euclidean Space2017
- U-Net: Convolutional Networks for Biomedical Image Segmentation2015
- Registration of images with affine geometric distortion based on Maximally Stable Extremal Regions and phase congruencyImage and Vision Computing, 2015
- Robust and Precise Registration of Oblique Images Based on Scale-Invariant Feature Transformation AlgorithmIEEE Geoscience and Remote Sensing Letters, 2012
- Effect size estimates: Current use, calculations, and interpretation.Journal of Experimental Psychology: General, 2012
- WEIGHTED PSEUDO-ALMOST PERIODIC SOLUTIONS FOR SOME ABSTRACT DIFFERENTIAL EQUATIONS WITH UNIFORM CONTINUITYBulletin of the Australian Mathematical Society, 2010
- ASIFT: A New Framework for Fully Affine Invariant Image ComparisonSIAM Journal on Imaging Sciences, 2009
- Seed Point Selection Method for Triangle Constrained Image Matching PropagationIEEE Geoscience and Remote Sensing Letters, 2006
- Distinctive Image Features from Scale-Invariant KeypointsInternational Journal of Computer Vision, 2004