Abstract
The objective of this study is to improve the robustness of homography estimation using deep learning for images superimposed with various types of disturbances. In conventional deep learning homography estimation, the original image and the image with perturbations and disturbances are simultaneously input to the model for estimation. The disadvantages of this approach are that the original image is affected by noise and the model itself is unclear. In this study, features are extracted separately for each of the two images, and a model is constructed based on ResNet using these features as input. In addition, when extracting the features of the perturbed and disturbed images, WaveCNet, which integrates the discrete wavelet transform into the CNN, is used to add pinpoint tolerance to the disturbance. The estimation accuracy of the homography matrix in the method proposed in this study shows improved accuracy in various noises. These results suggest that the proposed method is effective in reducing the effect of disturbance by extracting features robust to disturbance for each image.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Kanatani, K., Ohta, N.: Accuracy bounds and optimal computation of homography for image mosaicing applications. In: Proceedings of IEEE ICCV, vol. 1, pp. 73–78 (1999)
Chum, O., Pajdla, T., Sturmb, P.: The geometric error for homographies. Comput. Vis. Image Underst. 97, 86–102 (2005)
DeTone, D., Malisiewicz, T., Rabinovich, A.: Deep image homography estimation. arXiv:1606.03798 (2016)
Selvaraju, R.R., Cogswell, M., Das, A., Vedantam, R., Parikh, D., Batra, D.: Grad-CAM: visual explanations from deep networks via gradient-based localization. In: 2017 IEEE International Conference on Computer Vision (ICCV), pp. 618–626 (2017)
Nguyen, T., Chen, S.W., Shivakumar, S.S., Taylor, C.J., Kumar, V.: Unsupervised deep homography: a fast and robust homography estimation model. IEEE Robot. Autom. Lett. 3(3), 2346–2353 (2018)
Daubechies, I.: Ten Lectures on Wavelets, CBMS-NSF Regional Conference Series in Applied Mathematics 61. Society for Industrial and Applied Mathematics, Philadelphia (1992)
Mallat, S.G.: A theory for multiresolution signal decomposition: the wavelet representation. IEEE Trans. Pattern Anal. Mach. Intell. 11(7), 674–693 (1989)
Guo, T., Mousavi, H.S., Vu, T.H., Monga, V.: Deep wavelet prediction for image super-resolution. In: 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), pp. 1100–1109 (2017)
Azulay, A., Weiss, Y.: Why do deep convolutional networks generalize so poorly to small image transformations? arXiv preprint arXiv:1805.12177 (2018)
Li, Q., Shen, L., Guo, S., Lai, Z.: Wavelet integrated CNNs for noise-robust image classification. In: Proceedings of the IEEE/CVF Conference on Computer Vision Pattern Recognition (CVPR), pp. 7245–7254 (2020)
He, K., Zhang, X., Ren, S., Sun, J.: Deep residual learning for image recognition. arXiv:1512.03385 (2015)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2022 The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.
About this paper
Cite this paper
Yokono, M., Kamata, H. (2022). Improving the Accuracy of Homography Matrix Estimation for Disturbance Images Using Wavelet Integrated CNN. In: Fan, W., Zhang, L., Li, N., Song, X. (eds) Methods and Applications for Modeling and Simulation of Complex Systems. AsiaSim 2022. Communications in Computer and Information Science, vol 1712. Springer, Singapore. https://doi.org/10.1007/978-981-19-9198-1_42
Download citation
DOI: https://doi.org/10.1007/978-981-19-9198-1_42
Published:
Publisher Name: Springer, Singapore
Print ISBN: 978-981-19-9197-4
Online ISBN: 978-981-19-9198-1
eBook Packages: Computer ScienceComputer Science (R0)