In this work, we study the problem of line segment detection. In the recent studies of this problem, it was shown that
neural network-based methods are able to outperform classical algorithms in accuracy, yet high computational complexity
of neural networks limits their usage in real-world applications. In our work, we propose to fuse a neural network with
a classical algorithm to compensate their flaws, by (i) feeding into the network not only an image, but also classical
algorithm’s output and (ii) by combining their predictions. We expect these modifications to simplify the problem faced
by the neural network, resulting in higher accuracy in smaller network. Proposed method provides 0.72 and 0.66 F –
measure on Wireframe and York datasets respectively with 14 FPS on CPU.
artificial neural networks, line segment detection, contour detection, Unet, Canny
- Bandera A., Pérez-Lorenzo A., Bandera J.M., Sandoval F. Mean shift based clustering of hough domain for fast line segment detection. Pattern Recognition Letters. 2006. P. 578–586.
- Bista S.R., Giordano P.R., Chaumette F. Appearancebased indoor navigation by IBVS using line segments. IEEE Robotics and Automation Letters. 2016. P. 423–430.
- Burns J.B., Hanson A.R., Riseman E.M. Extracting straight lines. IEEE Transactions on Pattern Analysis and Machine Intelligence. 1986. P. 425–455.
- Canny J. A Computational approach to edge detection. IEEE Transactions on Pattern Analysis and Machine Intelligence. 1986. P. 679–698.
- Coughlan J.M., Yuille A.L. Manhattan world: orientation and outlier detection by bayesian inference. Neural Comput. 2003. P. 1063–1088.
- Duda R.O., Hart P.E. Use of the hough transformation to detect lines and curves in pictures. Commun. ACM. 1972. P. 11–15.
- Grompone von Gioi R., Jakubowicz J., Morel J.M., Randall G. LSD: A fast line segment detector with a false detection control. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2010. P. 722–732.
- Grompone von Gioi R., Jakubowicz J., Morel J.M., Randall G. LSD: A line segment detector. Image Processing On Lin. 2012. P. 35–55.
- Huang J., Wang Z., Liang H. Lane marking detection based on segments with upper and lower structure. International Journal of Pattern Recognition and Artificial Intelligence. 2020.
- Huang K., Wang Y., Zhou Z., Ding T., Gao S., Ma Y. Learning to parse wireframes in images of man-made environments. IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2018. P. 626–635.
- Kingma D.P., Ba J.L. Adam: A Method for stochastic optimization, 2014.
- Kiryati N., Eldar Y., Bruckstein A.M. A probabilistic hough transform. Pattern Recognition. 1991. P. 303–316.
- Krizhevsky A., Sutskever I., Hinton G.E. ImageNet classification with deep convolutional neural networks. Advances in Neural Information Processing Systems. 2012. P. 1097–1105.
- Lin Y., Silvia L., Pintea S. Deep hough-transform line priors, 2020.
- Ronneberger O., Fischer P., Tomas B. U-Net: convolutional networks for biomedical image segmentation. 2015.
- Xue N., Bai S., Wang F., Xia G., Wu T., Zhang L., Torr P. Learning regional attraction for line segment detection. IEEE transactions on pattern analysis and machine intelligence. 2019.
- Xue N., Wu T., Bai S., Wang F.D., Xia G.S., Zhang L. Holistically-attracted wireframe parsing, 2020.
- Yichao Z., Haozhi Q., Ma Y. End-to-end wireframe parsing. 2019.
- Zhukovsky A., Nikolaev D., Arlazarov V., Postnikov V., Polevoy D., Skoryukina N., Povolotsky M. Segments graph-based approach for document capture in a smartphone video stream. 14th IAPR International Conference on Document Analysis and Recognition (ICDAR). 2017. P. 337–342.