1 / 96

Computer and Robot Vision I

Computer and Robot Vision I. Chapter 8 The Facet Model. Presented by: 蘇唯誠 r02902114@ntu.edu.tw 指導教授 : 傅楸善 博士. 8.1 Introduction. facet model: image as continuum or piecewise continuous intensity surface observed digital image: noisy discretized sampling of distorted version Why?

hasad
Download Presentation

Computer and Robot Vision I

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Computer and Robot Vision I Chapter 8 The Facet Model Presented by: 蘇唯誠 r02902114@ntu.edu.tw 指導教授: 傅楸善 博士 Digital Camera and Computer Vision Laboratory Department of Computer Science and Information Engineering National Taiwan University, Taipei, Taiwan, R.O.C.

  2. 8.1 Introduction • facet model: image as continuum or piecewise continuous intensity surface • observed digital image: noisy discretized sampling of distorted version • Why? • Before processing images, must define what the conditioning means with respect to the underlying gray level intensity surface. • Conditioning estimates informative pattern. DC & CV Lab. CSIE NTU

  3. 8.1 Introduction DC & CV Lab. CSIE NTU

  4. 8.1 Introduction • general forms: 1. piecewise constant (flat facet model), ideal region: constant gray level 2. piecewise linear (sloped facet model), ideal region: sloped plane gray level 3. piecewise quadratic, gray level surface: bivariate quadratic 4. piecewise cubic, gray level surface: cubic surfaces DC & CV Lab. CSIE NTU

  5. 8.2 Relative Maxima • a simple labeling application • Detect and locate all relative maxima • relative maxima: • first derivative zero • second derivative negative DC & CV Lab. CSIE NTU

  6. 8.2 Relative Maxima • One-dimensional observation sequence ,…… • We can least-squares fit a quadratic function , where , to each group of 2k+1 successive observations DC & CV Lab. CSIE NTU

  7. 8.2 Relative Maxima • The squared fitting error for the group of 2k+1 can be expressed by DC & CV Lab. CSIE NTU

  8. 8.2 Relative Maxima • Taking partial derivatives of DC & CV Lab. CSIE NTU

  9. 8.2 Relative Maxima • Assume k = 1, setting these partial derivatives to zero DC & CV Lab. CSIE NTU

  10. 8.2 Relative Maxima • The quadratic has relative extrema at • The extremum is a relative maximum when DC & CV Lab. CSIE NTU

  11. 8.2 Relative Maxima DC & CV Lab. CSIE NTU

  12. 8.3 Sloped Facet Parameter and Error Estimation • Least-squares procedure: to estimate sloped facet parameter, noise variance • image function • Where is a random variable indexed on , which represents noise. • and that the noise for any two pixels is independent. DC & CV Lab. CSIE NTU

  13. 8.3 Sloped Facet Parameter and Error Estimation • determine parameters that minimize the sum of the squared differences DC & CV Lab. CSIE NTU

  14. 8.3 Sloped Facet Parameter and Error Estimation • Taking the partial derivatives of and setting them to zero DC & CV Lab. CSIE NTU

  15. 8.3 Sloped Facet Parameter and Error Estimation • Without loss of generality, we choose our coordinate RxC so that the center of the neighborhood RxC has coordinates (0, 0) • The symmetry in the chosen coordinate system leads to DC & CV Lab. CSIE NTU

  16. 8.3 Sloped Facet Parameter and Error Estimation • Solving for , , and DC & CV Lab. CSIE NTU

  17. 8.3 Sloped Facet Parameter and Error Estimation • Replacing by DC & CV Lab. CSIE NTU

  18. 8.3 Sloped Facet Parameter and Error Estimation • is noise having mean 0 and variance DC & CV Lab. CSIE NTU

  19. 8.3 Sloped Facet Parameter and Error Estimation • Examining the squared error residual DC & CV Lab. CSIE NTU

  20. 8.3 Sloped Facet Parameter and Error Estimation • Using • We obtain DC & CV Lab. CSIE NTU

  21. 8.3 Sloped Facet Parameter and Error Estimation • chi-distribution • Assume , , …… , are independent variable , where • is distributed according to the chi-squared distribution with n degrees of freedom. DC & CV Lab. CSIE NTU

  22. 8.3 Sloped Facet Parameter and Error Estimation • is a chi-squared distribution with degrees of freedom. DC & CV Lab. CSIE NTU

  23. 8.3 Sloped Facet Parameter and Error Estimation • Because , , and are independent normal distributions • is a chi-squared distribution with 3 degrees of freedom. DC & CV Lab. CSIE NTU

  24. 8.3 Sloped Facet Parameter and Error Estimation • is distributed as a chi-squared variate with degrees of freedom. • This means that can be used as an unbiased estimator for DC & CV Lab. CSIE NTU

  25. 8.4 Facet-Based Peak Noise Removal • peak noise pixel: gray level intensity significantly differs from neighbors • (a) peak noise pixel, (b) not DC & CV Lab. CSIE NTU

  26. 8.4 Facet-Based Peak Noise Removal • Center-deleted neighborhood • Where is assumed to be independent additive Gaussian noise having mean 0 and variance DC & CV Lab. CSIE NTU

  27. 8.4 Facet-Based Peak Noise Removal • The minimizing , , and are given by DC & CV Lab. CSIE NTU

  28. 8.4 Facet-Based Peak Noise Removal • Under the hypothesis that is not peak noise, has a Gaussian distribution with mean 0 and variance • Hence has mean 0 and variance 1 DC & CV Lab. CSIE NTU

  29. 8.4 Facet-Based Peak Noise Removal • We have already obtained that has a chi-squared distribution with degrees of freedom. • Hence has a t-distribution with degrees of freedom DC & CV Lab. CSIE NTU

  30. 8.4 Facet-Based Peak Noise Removal • t-distribution • (0,1), , and are independent. • is a t-distribution with n degrees of freedom, denoted as DC & CV Lab. CSIE NTU

  31. 8.4 Facet-Based Peak Noise Removal • Let be the number satisfying • is a threshold. DC & CV Lab. CSIE NTU

  32. 8.4 Facet-Based Peak Noise Removal • If , the hypothesis of the equality of and is rejected, and the output value for the center pixel is given by . • If , the hypothesis of the equality of and is not rejected, and the output value for the center pixel is given by . DC & CV Lab. CSIE NTU

  33. 8.4 Facet-Based Peak Noise Removal • The reasonable value for p is 0.05 to 0.01 DC & CV Lab. CSIE NTU

  34. 8.5 Iterated Facet Model • The iterated model for ideal image assume that the spatial of the image can be partitioned into connected regions called facets, each of which satisfies certain gray level and shape constraints. DC & CV Lab. CSIE NTU

  35. 8.5 Iterated Facet Model • gray level constraint: • gray levels in each facet must be a polynomial function of row-column coordinates • shape constraint: • each facet must be sufficiently smooth in shape DC & CV Lab. CSIE NTU

  36. 8.5 Iterated Facet Model • Each pixel in image is contained in different blocks. • Each block fits a polynomial model. • Set the output gray value to be that gray value fitted by the block with smallest error variance. DC & CV Lab. CSIE NTU

  37. 8.6 Gradient-Based Facet Edge Detection • gradient-based facet edge detection: high values in first partial derivative (sharp discontinuities) DC & CV Lab. CSIE NTU

  38. 8.6 Gradient-Based Facet Edge Detection • When a neighborhood is fitted with the sloped facet model , , a gradient magnitude of will result. • Use the estimated gradient as the basis for edge detection. DC & CV Lab. CSIE NTU

  39. 8.6 Gradient-Based Facet Edge Detection • , are independent DC & CV Lab. CSIE NTU

  40. 8.6 Gradient-Based Facet Edge Detection • It follows that to test the hypothesis of no edge under the assumption that , we use the statistic G: DC & CV Lab. CSIE NTU

  41. 8.6 Gradient-Based Facet Edge Detection • The false-alarm rate is the conditional probability that the edge detector classifies a pixel as an edge given that the pixel is not a edge. • Suppose the false-alarm rate is to be held to 1%. Then since , the threshold we used must be at least 9.21 DC & CV Lab. CSIE NTU

  42. 8.7 Bayesian Approach to Gradient Edge Detection • The Bayesian approach to the decision of whether or not an observed gradient magnitude G is statistically significant and therefore participates in some edge is to decide there is an edge (statistically significant gradient) when, • : given gradient magnitude conditional probability of edge • : given gradient magnitude conditional probability of nonedge DC & CV Lab. CSIE NTU

  43. 8.7 Bayesian Approach to Gradient Edge Detection • Hence a decision for edge is made whenever • is known to be the density function of a variate. DC & CV Lab. CSIE NTU

  44. 8.7 Bayesian Approach to Gradient Edge Detection • Infer from DC & CV Lab. CSIE NTU

  45. 8.7 Bayesian Approach to Gradient Edge Detection • The threshold t for G is determined as that value t for which • Implies that the threshold t must satisfy that • is a user-specified prior probability of nonedge, 0.9 to 0.95 are reasonable. DC & CV Lab. CSIE NTU

  46. 8.8 Zero-Crossing Edge Detector • compare • gradient edge detector: looks for high values of first derivatives • zero-crossing edge detector: looks for relative maxima in first derivative • zero-crossing: pixel as edge if zero crossing of second directional derivative underlying gray level intensity function f takes the form DC & CV Lab. CSIE NTU

  47. 8.8.1 Discrete Orthogonal Polynomials • The underlying functions from which the directional derivative are computed are easy to represent as linear combinations of the polynomials in any polynomial basis set. • basis set: the discrete orthogonal polynomials DC & CV Lab. CSIE NTU

  48. 8.8.1 Discrete Orthogonal Polynomials • discrete orthogonal polynomial basis set of size N: polynomials deg. 0..N - 1 • discrete Chebyshev polynomials: these unique polynomials DC & CV Lab. CSIE NTU

  49. 8.8.1 Discrete Orthogonal Polynomials (cont’) • discrete orthogonal polynomials can be recursively generated , DC & CV Lab. CSIE NTU

  50. 8.8.2 Two-Dimensional Discrete Orthogonal Polynomials • 2-D discrete orthogonal polynomials creatable from tensor products of 1Dfrom above equations _ r DC & CV Lab. CSIE NTU

More Related