Natural Language Understanding Wiki

Types of edges[]

Classified based on appearance[]

  • Edges defined based on brightness or color. Taxonomy based on
    • Step/jump edges: the image intensity abruptly changes from one value to one side of thediscontinuity to a different value on the opposite side
    • Ramp edges: a step edge where the intensity change is not instantaneous but occuroverafinite distance.
    • Ridge/crease edges: the image intensity abruptly changes value but then returns to thestarting value within some short distance (generated usually by lines).
    • Roof edges: a ridge edge where the intensity change is not instantaneous but occuroverafinite distance (generated usually by the intersection of surfaces)."
  • Edges defined based on texture: see Martin et al. (2003)[1]

Classified based on origin[]

According to Koschan & Abidi (2005)[2]: "shadow edges, reflectance edges, orientation edges, occlusion edges, and specular edges."

Edge detection methods[]

TODO: Check He, J., Zhang, S., Yang, M., Shan, Y., & Huang, T. (2019). Bi-directional cascade network for perceptual edge detection. ArXiv, 3828–3837.

Most edge detection methods can be categorized into three groups, i.e., traditional edge oper- ators, learning based methods, and the recent deep learn- ing, respectively. Traditional edge operators [22, 4, 45, 34] detect edges by finding sudden changes in intensity, col- or, texture, etc. Learning based methods spot edges by u- tilizing supervised models and hand-crafted features. For example, Doll´ar et al. [10] propose structured edge which jointly learns the clustering of groundtruth edges and the mapping of image patch to clustered token. Deep learning based methods use CNN to extract multi-level hierarchical features. Bertasius et al. [2] employ CNN to generate fea- tures of candidate contour points. Xie et al. [49] propose an end-to-end detection model that leverages the output- s from different intermediate layers with skip-connections. Liu et al. [30] further learn richer deep representations by concatenating features derived from all convolutional lay- ers. Xu et al. [51] introduce a hierarchical deep model to extract multi-scale features and a gated conditional random field to fuse them.

TODO: any other way to classify methods?

Early methods[]

Data-driven methods with hand-crafted features[]

  • Pb features?[1]
  • gPb features?[7]
  • Sketch tokens?[8]
  • Structured forests[9]

Neural networks[]

  • Encoder-decoder:
    • Srinivasan et al. 1994[10]: encoder: 16 neurons with 15x15 receptive field, decoder: two outputs representing the components of an edge vector, training: one (!) synthetic image with edges of various strengths and orientations, evaluation on two images.
  • "Dynamic NN"?[11]: 3 layers
  • "Hopfield network?"[12]
  • N4-fields?[13]
  • cGAN[14]
  • HED?


Object detection[]

Method Edge detection technique Neural edge detection? Comment
Ferrari et al. (2008)[15] N

Object proposal[]

Method Edge detection technique Neural edge detection? Comment
BING[16] Normed gradient N
Edge boxes[17] Structured Edge detector[9] (decision tree) N Count edges that a bounding box whole contains and those that it doesn't

Image segmentation[]

Method Edge detection technique Neural edge detection? Comment
Multiscale Combinatorial Grouping[18] Structured Edge detector[9] (decision tree) N

See also[]


  1. 1.0 1.1 D. R. Martin, C. C. Fowlkes, and J. Malik. Learning to detect natural image boundaries using local brightness, color, and texture cues. IEEE TPAMI, 26(5):530–549, 2004.
  2. Koschan, A., & Abidi, M. (2005). Detection and classification of edges in color images [a review of vector-valued techniques]. Signal Processing Magazine, 22(1), 64–73.
  3. G. S. Robinson. Color edge detection. Optical Engineering, 16(5):165479–165479, 1977.
  4. Roberts, L. G. (1963). Machine perception of three-dimensional solids, (January 1963), 82.
  5. Sobel, I. (2014). History and Definition of Sobel Operator.
  6. Canny, J. (1986). A Computational Approach to Edge Detection. IEEE Transactions on Pattern Analysis and Machine Intelligence, PAMI-8(6), 679–698.
  7. P.Arbela ́ez,M.Maire,C.Fowlkes,andJ.Malik.Contourde- tection and hierarchical image segmentation. IEEE TPAMI, 33(5):898–916, 2011.
  8. J. J. Lim, C. L. Zitnick, and P. Dolla ́r. Sketch tokens: A learned mid-level representation for contour and object de- tection. In IEEE CVPR, pages 3158–3165, 2013.
  9. 9.0 9.1 9.2 Dollar, P., & Zitnick, C. L. (2013). Structured forests for fast edge detection. Proceedings of the IEEE International Conference on Computer Vision, 1841–1848.
  10. Srinivasan, V., Bhatia, P., & Ong, S. H. (1994). Edge detection using a neural network. Pattern Recognition, 27(12), 1653–1662.
  11. Chandrasekaran, V., Palaniswami, M., & Caelli, T. M. (1996). Range image segmentation by dynamic neural network architecture. Pattern Recognition, 29(2), 315–329.
  12. C.-T. Tsai, Y.-N. Sun, P.-C. Chung et al., Endocardial boundary detection using a neural network, Pattern Recognition 26 (7) (1993) 1057–1068.
  13. Y. Ganin and V. Lempitsky. N -Fields: Neural network nearest neighbor fields for image transforms. In ACCV, pages 536–551. Springer, 2014
  14. Zeng, Z., Yu, Y. K., & Hong Wong, K. (2019). Adversarial network for edge detection. 2018 Joint 7th International Conference on Informatics, Electronics and Vision and 2nd International Conference on Imaging, Vision and Pattern Recognition, ICIEV-IVPR 2018, 19–23.
  15. Ferrari, V., Fevrier, L., Jurie, F., & Schmid, C. (2008). Groups of adjacent contour segments for object detection. IEEE Transactions on Pattern Analysis and Machine Intelligence, 30(1), 36–51.
  16. Cheng, M. M., Liu, Y., Lin, W. Y., Zhang, Z., Rosin, P. L., & Torr, P. H. S. (2019). BING: Binarized normed gradients for objectness estimation at 300fps. Computational Visual Media, 5(1), 3–20.
  17. Zitnick, C. L., & Dollár, P. (2014). Edge boxes: Locating object proposals from edges. Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 8693 LNCS(PART 5), 391–405.
  18. Arbeláez, P., Pont-Tuset, J., Barron, J., Marques, F., & Malik, J. (2014). Multiscale combinatorial grouping. Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 500, 328–335.