Deep Hough-Transform Line Priors

Classical work on line segment detection is knowledge-based; it uses carefully designed geometric priors using either image gradients, pixel groupings, or Hough transform variants. Instead, current deep learning methods do away with all prior knowledge and replace priors by training deep networks on large manually annotated datasets. Here, we reduce the dependency on labeled data by building on the classic knowledge-based priors while using deep networks to learn features. We add line priors through a trainable Hough transform block into a deep network. Hough transform provides the prior knowledge about global line parameterizations, while the convolutional layers can learn the local gradient-like line features. On the Wireframe (ShanghaiTech) and York Urban datasets we show that adding prior knowledge improves data efficiency as line priors no longer need to be learned from data. Keywords: Hough transform; global line prior, line segment detection.

[1]  Allen R. Hanson,et al.  Extracting Straight Lines , 1986, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[2]  Bok-Suk Shin,et al.  Accurate and Robust Line Segment Extraction Using Minimum Entropy With Hough Transform , 2015, IEEE Transactions on Image Processing.

[3]  Rafael Grompone von Gioi,et al.  On Straight Line Segment Detection , 2008, Journal of Mathematical Imaging and Vision.

[4]  Harry Shum,et al.  Image vectorization using optimized gradient meshes , 2007, ACM Trans. Graph..

[5]  Mohan M. Trivedi,et al.  Efficient Lane and Vehicle Detection with Integrated Synergies (ELVIS) , 2014, 2014 IEEE Conference on Computer Vision and Pattern Recognition Workshops.

[6]  Li-Yi Wei,et al.  Learning to Reconstruct 3D Manhattan Wireframes From a Single Image , 2019, 2019 IEEE/CVF International Conference on Computer Vision (ICCV).

[7]  Lorenzo Porzi,et al.  A Deeply-Supervised Deconvolutional Network for Horizon Line Detection , 2016, ACM Multimedia.

[8]  Matthew Richardson,et al.  Do Deep Convolutional Nets Really Need to be Deep and Convolutional? , 2016, ICLR.

[9]  Bok-Suk Shin,et al.  A statistical method for line segment detection , 2015, Comput. Vis. Image Underst..

[10]  In So Kweon,et al.  VPGNet: Vanishing Point Guided Network for Lane and Road Marking Detection and Recognition , 2017, 2017 IEEE International Conference on Computer Vision (ICCV).

[11]  Trevor Darrell,et al.  Blurring the Line Between Structure and Learning to Optimize and Adapt Receptive Fields , 2019, ArXiv.

[12]  Pierre Gurdjos,et al.  A Parameterless Line Segment and Elliptical Arc Detector with Enhanced Ellipse Fitting , 2012, ECCV.

[13]  Marie-Odile Berger,et al.  A-Contrario Horizon-First Vanishing Point Detection Using Second-Order Grouping Laws , 2018, ECCV.

[14]  Cristina Campi,et al.  Geometry of the Hough Transforms with Applications to Synthetic Data , 2019, ArXiv.

[15]  Samuel Kosolapov,et al.  Horizon Line Detection in Marine Images: Which Method to Choose? , 2013 .

[16]  Jiri Matas,et al.  Robust Detection of Lines Using the Progressive Probabilistic Hough Transform , 2000, Comput. Vis. Image Underst..

[17]  Leonidas J. Guibas,et al.  Deep Hough Voting for 3D Object Detection in Point Clouds , 2019, 2019 IEEE/CVF International Conference on Computer Vision (ICCV).

[18]  Boris Katz,et al.  ObjectNet: A large-scale bias-controlled dataset for pushing the limits of object recognition models , 2019, NeurIPS.

[19]  Jian Sun,et al.  Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification , 2015, 2015 IEEE International Conference on Computer Vision (ICCV).

[20]  Dmitry P. Nikolaev,et al.  Hough Transform: Underestimated Tool In The Computer Vision Field , 2008 .

[21]  Subramaniam Ganesan,et al.  Complete description of multiple line segments using the Hough transform , 1998, Image Vis. Comput..

[22]  Yue Zhang,et al.  X-LineNet: Detecting Aircraft in Remote Sensing Images by a Pair of Intersecting Line Segments , 2019, IEEE Transactions on Geoscience and Remote Sensing.

[23]  Jitendra Malik,et al.  Using contours to detect and localize junctions in natural images , 2008, 2008 IEEE Conference on Computer Vision and Pattern Recognition.

[24]  Jianhua Ma,et al.  Radon inversion via deep learning , 2018, Medical Imaging.

[25]  Ju Jia Zou,et al.  Cartoon image vectorization based on shape subdivision , 2001, Proceedings. Computer Graphics International 2001.

[26]  Dmitry Nikolaev,et al.  HoughNet: Neural Network Architecture for Vanishing Points Detection , 2019, 2019 International Conference on Document Analysis and Recognition (ICDAR).

[27]  Pedro M. Q. Aguiar,et al.  Connectivity-Enforcing Hough Transform for the Robust Extraction of Line Segments , 2011, IEEE Transactions on Image Processing.

[28]  Jen Beatty The Radon Transform and the Mathematics of Medical Imaging , 2012 .

[29]  Scott Workman,et al.  Horizon Lines in the Wild , 2016, BMVC.

[30]  P. Toft The Radon Transform - Theory and Implementation , 1996 .

[31]  Kun Huang,et al.  Learning to Parse Wireframes in Images of Man-Made Environments , 2018, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.

[32]  Alan Yuille,et al.  A Novel Linelet-Based Representation for Line Segment Detection , 2018, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[33]  Stéphane Mallat,et al.  Invariant Scattering Convolution Networks , 2012, IEEE transactions on pattern analysis and machine intelligence.

[34]  Ronen Lerner,et al.  Recent progress in road and lane detection: a survey , 2012, Machine Vision and Applications.

[35]  Da-Zheng Feng,et al.  Funnel Transform for Straight Line Detection , 2019, ArXiv.

[36]  James H. Elder,et al.  MCMLSD: A Dynamic Programming Approach to Line Segment Detection , 2017, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[37]  Yoshihisa Shinagawa,et al.  Accurate and robust line segment extraction by analyzing distribution around peaks in Hough space , 2003, Comput. Vis. Image Underst..

[38]  Jitendra Malik,et al.  Learning to detect natural image boundaries using local brightness, color, and texture cues , 2004, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[39]  Ivan Sosnovik,et al.  Scale-Equivariant Steerable Networks , 2020, ICLR.

[40]  Robert Jenssen,et al.  LS-Net: Fast Single-Shot Line-Segment Detector , 2021, Mach. Vis. Appl..

[41]  Gui-Song Xia,et al.  Holistically-Attracted Wireframe Parsing , 2020, 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[42]  Jianwei Niu,et al.  Robust Lane Detection using Two-stage Feature Extraction with Curve Fitting , 2016, Pattern Recognit..

[43]  Richard O. Duda,et al.  Use of the Hough transformation to detect lines and curves in pictures , 1972, CACM.

[44]  Jean Ponce,et al.  Hyperpixel Flow: Semantic Correspondence With Multi-Layer Neural Features , 2019, 2019 IEEE/CVF International Conference on Computer Vision (ICCV).

[45]  Shenghua Gao,et al.  PPGNet: Learning Point-Pair Graph for Line Segment Detection , 2019, 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[46]  Arnold W. M. Smeulders,et al.  Structured Receptive Fields in CNNs , 2016, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[47]  Michael S. Bernstein,et al.  ImageNet Large Scale Visual Recognition Challenge , 2014, International Journal of Computer Vision.

[48]  Donsub Rim Exact and Fast Inversion of the Approximate Discrete Radon Transform , 2020, Appl. Math. Lett..

[49]  Gui-Song Xia,et al.  Learning Attraction Field Representation for Robust Line Segment Detection , 2018, 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[50]  Jian Sun,et al.  Deep Residual Learning for Image Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[51]  James H. Elder,et al.  Efficient Edge-Based Methods for Estimating Manhattan Frames in Urban Imagery , 2008, ECCV.

[52]  Rafael Grompone von Gioi,et al.  LSD: A Fast Line Segment Detector with a False Detection Control , 2010, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[53]  Jan C. van Gemert,et al.  On Translation Invariance in CNNs: Convolutional Layers can Exploit Absolute Spatial Location , 2020, CVPR.

[54]  Minh N. Do,et al.  The finite ridgelet transform for image representation , 2003, IEEE Trans. Image Process..

[55]  Yi Ma,et al.  End-to-End Wireframe Parsing , 2019, 2019 IEEE/CVF International Conference on Computer Vision (ICCV).