Natural Language Understanding Wiki
Advertisement

Types of edges[]

Classified based on appearance[]

  • Edges defined based on brightness or color. Taxonomy based on https://www.cse.unr.edu/~bebis/CS791E/Notes/EdgeDetection.pdf
    • 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. https://doi.org/10.1109/tpami.2020.3007074

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?

Application[]

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[]

References[]

  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. https://doi.org/10.1109/TPAMI.1986.4767851
  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. https://doi.org/10.1109/ICCV.2013.231
  10. Srinivasan, V., Bhatia, P., & Ong, S. H. (1994). Edge detection using a neural network. Pattern Recognition, 27(12), 1653–1662. https://doi.org/10.1016/0031-3203(94)90084-1
  11. Chandrasekaran, V., Palaniswami, M., & Caelli, T. M. (1996). Range image segmentation by dynamic neural network architecture. Pattern Recognition, 29(2), 315–329. https://doi.org/10.1016/0031-3203(95)00038-0
  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. https://doi.org/10.1109/ICIEV.2018.8641005
  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. https://doi.org/10.1109/TPAMI.2007.1144
  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. https://doi.org/10.1007/s41095-018-0120-1
  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. https://doi.org/10.1007/978-3-319-10602-1_26
  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. https://doi.org/10.1109/CVPR.2014.49
Advertisement