Presentation is loading. Please wait.

Presentation is loading. Please wait.

Hough Transform on the GPU Presented by Javier Garcia November 29, 2010.

Similar presentations


Presentation on theme: "Hough Transform on the GPU Presented by Javier Garcia November 29, 2010."— Presentation transcript:

1 Hough Transform on the GPU Presented by Javier Garcia November 29, 2010

2 Why Hough transform? Sound wave problem ◦ Too complex Image processing ◦ 2 for 1

3 What is it? Feature extraction technique Used to detect lines (or other curves) in images

4 How does it work? It finds curves in images using a voting technique The voting takes place in a parameter space or Hough space Object (curve) candidates are obtained as local maxima

5 Original algorithm Used slope intercept form for lines ◦ y = ax + b ◦ b = -xa + y Slope of vertical lines could be a problem Quantize parameter space (a, b): P[a min,..., a max ][b min,..., b max ] (accumulator array) For each edge point (x, y) For(a = a min ; a ≤ a max ; a++) { b = − xa + y; P[a][b]++; }

6 Original algorithm (cont’d) Find local maxima in P[a][b] If P[a j ][b k ]=M, then M points lie on the line y = a j x + b k Hough transformation in slope-intercept form: http://www.rob.cs.tu-bs.de/content/04- teaching/06-interactive/Hough.html

7 Generalized algorithm Use polar coordinates for line: ρ = x cos θ + y sin θ Quantize parameter space ( ρ, θ ): P[ ρ min,..., ρ max ][ θ min,..., θ max ] (accumulator array) For each edge point (x, y) For( θ = θ min ; θ ≤ θ max ; θ ++) { ρ = xcos( θ ) + ysin( θ ) ; P[ ρ ][ θ ]++; }

8 Generalized algorithm (cont’d) Find local maxima in P[ ρ ][ θ ] Using generalized algorithm, any curve can be extracted as long as it can be described by an equation analytically ◦ The equation for a circle is: r 2 = (x - x 0 ) 2 + (y - y 0 ) 2 Accumulator dimensions are equal to the unknown parameters Hough transformation in normal form: http://www.rob.cs.tu-bs.de/content/04- teaching/06-interactive/HNF.html

9 Quantization

10 My approach Implement the generalized Hough transform on the CPU and GPU Use both versions to extract lines from images Vary quantization ◦ Finer should give better results but increases space and time requirements ◦ Coarser is better for noise tolerance Finally, extract curves other than lines (time permitting)

11 Expected results I expect the GPU version to perform faster than the CPU version (of course) Fine quantization should be slower than coarse I will be presenting the execution times with the varied parameters for the final

12 Questions?


Download ppt "Hough Transform on the GPU Presented by Javier Garcia November 29, 2010."

Similar presentations


Ads by Google