Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one...
Transcript of Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one...
![Page 1: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/1.jpg)
Lecture 2: Edge detection
CS5670: Computer VisionNoah Snavely
From Sandlot Science
![Page 2: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/2.jpg)
Announcements
• Project 1 (Hybrid Images) is now on the course webpage (see “Projects” link)– Due Wednesday, Feb 15, by 11:59pm
– Artifact due Friday, Feb 15, by 11:59pm
– To be done individually
– Voting system for favorite artifacts (with small amount of extra credit)
– Files due by CMS
– We will provide a course VM for you to run the assignments
![Page 3: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/3.jpg)
Project 1: Hybrid Images
![Page 4: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/4.jpg)
Project 1 Demo
![Page 5: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/5.jpg)
Edge detection
• Convert a 2D image into a set of curves
– Extracts salient features of the scene
– More compact than pixels
![Page 6: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/6.jpg)
Origin of Edges
• Edges are caused by a variety of factors
depth discontinuity
surface color discontinuity
illumination discontinuity
surface normal discontinuity
![Page 7: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/7.jpg)
Images as functions…
• Edges look like steep cliffs
![Page 8: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/8.jpg)
Characterizing edges
• An edge is a place of rapid change in the image intensity function
imageintensity function
(along horizontal scanline) first derivative
edges correspond toextrema of derivativeSource: L. Lazebnik
![Page 9: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/9.jpg)
• How can we differentiate a digital image F[x,y]?
– Option 1: reconstruct a continuous image, f, then compute the derivative
– Option 2: take discrete derivative (finite difference)
1 -1
How would you implement this as a linear filter?
Image derivatives
-1
1
: :
Source: S. Seitz
![Page 10: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/10.jpg)
The gradient points in the direction of most rapid increase in intensity
Image gradient
• The gradient of an image:
The edge strength is given by the gradient magnitude:
The gradient direction is given by:
• how does this relate to the direction of the edge?Source: Steve Seitz
![Page 11: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/11.jpg)
Image gradient
Source: L. Lazebnik
![Page 12: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/12.jpg)
Effects of noise
Where is the edge?Source: S. Seitz
Noisy input image
![Page 13: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/13.jpg)
Solution: smooth first
f
h
f * h
Source: S. Seitz
To find edges, look for peaks in
![Page 14: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/14.jpg)
• Differentiation is convolution, and convolution is associative:
• This saves us one operation:
Associative property of convolution
f
Source: S. Seitz
![Page 15: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/15.jpg)
The 1D Gaussian and its derivatives
![Page 16: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/16.jpg)
2D edge detection filters
Gaussianderivative of Gaussian (x)
![Page 17: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/17.jpg)
Derivative of Gaussian filter
x-direction y-direction
![Page 18: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/18.jpg)
The Sobel operator
• Common approximation of derivative of Gaussian
-1 0 1
-2 0 2
-1 0 1
1 2 1
0 0 0
-1 -2 -1
• The standard defn. of the Sobel operator omits the 1/8 term– doesn’t make a difference for edge detection
– the 1/8 term is needed to get the right gradient magnitude
![Page 19: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/19.jpg)
Sobel operator: example
Source: Wikipedia
![Page 20: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/20.jpg)
![Page 21: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/21.jpg)
Example
• original image (Lena)
![Page 22: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/22.jpg)
Finding edges
gradient magnitude
![Page 23: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/23.jpg)
thresholding
Finding edges
where is the edge?
![Page 24: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/24.jpg)
Get Orientation at Each Pixel
• Threshold at minimum level
• Get orientation
theta = atan2(gy, gx)
![Page 25: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/25.jpg)
• Check if pixel is local maximum along gradient direction
– requires interpolating pixels p and r
Non-maximum supression
![Page 26: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/26.jpg)
Before Non-max Suppression
![Page 27: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/27.jpg)
After Non-max Suppression
![Page 28: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/28.jpg)
thresholding
Finding edges
![Page 29: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/29.jpg)
thinning(non-maximum suppression)
Finding edges
![Page 30: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/30.jpg)
Canny edge detector
1. Filter image with derivative of Gaussian
2. Find magnitude and orientation of gradient
3. Non-maximum suppression
4. Linking and thresholding (hysteresis):– Define two thresholds: low and high
– Use the high threshold to start edge curves and the low threshold to continue them
Source: D. Lowe, L. Fei-Fei
MATLAB: edge(image,‘canny’)
![Page 31: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/31.jpg)
Canny edge detector
• Still one of the most widely used edge detectors in computer vision
• Depends on several parameters:
J. Canny, A Computational Approach To Edge Detection, IEEE Trans. Pattern Analysis and Machine Intelligence, 8:679-714, 1986.
: width of the Gaussian blur
high thresholdlow threshold
![Page 32: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/32.jpg)
Canny edge detector
Canny with Canny with original
• The choice of depends on desired behavior
– large detects “large-scale” edges
– small detects fine edges
Source: S. Seitz
![Page 33: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/33.jpg)
Scale space (Witkin 83)
• Properties of scale space (w/ Gaussian smoothing)
– edge position may shift with increasing scale ()
– two edges may merge with increasing scale
– an edge may not split into two with increasing scale
larger
Gaussian filtered signal
first derivative peaks
![Page 34: Lecture 2: Edge detection - Cornell University · 2017-02-02 · Canny edge detector • Still one of the most widely used edge detectors in computer vision • Depends on several](https://reader034.fdocuments.us/reader034/viewer/2022050105/5f43502bb529c74c64766ec9/html5/thumbnails/34.jpg)
Questions?