Edge Detection

• Identify sudden changes in image

  • most semantic info from an image can be encoded in edges
  • more compact than pixels

• edges allow us to

  • extract info
  • object recognition
  • recover geometry and viewpoint

• Where surfaces begin and end help analyze the world; discontinuity

  • surface normal
  • depth
  • surface color
  • illumination

• Image gradient

• when you take gradient of intensity, you get peaks

---

• Criteria for “optimal”

  • good detection: minimize probability of false positives (edges caused by noise), and false negatives (missing real edges)
  • good localization: edges detected should be as close as possible to true edges
    • i.e., blurred edges make them thicker
  • Single response: return one point only for each true edge point