1 / 78

Computer and Robot Vision I

Computer and Robot Vision I. Chapter 11 Arc Extraction and Segmentation. Presented by: 傅楸善 & 何育哲 0937 960 615 r94922131@ntu.edu.tw 指導教授 : 傅楸善 博士. 11.1 Introduction. edge detection: labels each pixel as edge or no edge

richardmoss
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 11 Arc Extraction and Segmentation Presented by: 傅楸善 & 何育哲 0937 960 615 r94922131@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. 11.1 Introduction • edge detection: labels each pixel as edge or no edge • additional properties of edge: direction, gradient magnitude, contrast • edge grouping: edge pixels in same region boundary grouped together • edge detector: typically produces short linear disjointed edge segments DC & CV Lab. CSIE NTU

  3. 11.1 Introduction • each edge segment: has an orientation and a position • edge segments: of little use until aggregated into extended edges • global edge aggregation: upon proximity, relative orientation, contrast, … • local edge aggregation: extends edges by seeking compatible neighbor edges DC & CV Lab. CSIE NTU

  4. 11.1 Introduction • global methods: can incorporate domain knowledge into cost functions • global methods: more robust to noisy edge data • global methods: bought at expense of relatively high computational cost • aggregation into extended edges and parametric description: new research area • extended edges: like other curves can be described at multiple scales DC & CV Lab. CSIE NTU

  5. 11.1 Introduction • Fig. 3.18(a) DC & CV Lab. CSIE NTU

  6. 11.1 Introduction • Fig. 3.18(b) DC & CV Lab. CSIE NTU

  7. 11.2 Extracting Boundary Pixels from a Segmented Image • after segmentation/ connected components: region boundary may be extracted • border algorithm: to extract all region boundaries in left-right, top-bottom DC & CV Lab. CSIE NTU

  8. 11.2.1 Concepts and Data Structures • input image: symbolic image whose pixel values denote region labels • current regions: borders partially scanned but not yet output • past regions: completely scanned and their borders output DC & CV Lab. CSIE NTU

  9. 11.2.1 Concepts and Data Structures • future regions: not yet been reached by the scan • at most 2 * number_of_columns region labels may be active at a time • hash table: used to allow rapid access to chains of a region • each chain: linked list of pixel positions grow-able from beginning to end DC & CV Lab. CSIE NTU

  10. 11.2.2 Border-Tracking Algorithm DC & CV Lab. CSIE NTU

  11. 11.2.2 Border-Tracking Algorithm DC & CV Lab. CSIE NTU

  12. 11.3 Linking One-Pixel-Wide Edges or Lines • segments: may consist of line, endpoint, corner, junctions DC & CV Lab. CSIE NTU

  13. 11.3 Linking One-Pixel-Wide Edges or Lines DC & CV Lab. CSIE NTU

  14. 11.3 Linking One-Pixel-Wide Edges or Lines • algorithm to track segments like these has to be concerned with • starting a new segment • adding an interior pixel to a segment • ending a segment • finding a junction • finding a corner DC & CV Lab. CSIE NTU

  15. 11.3 Linking One-Pixel-Wide Edges or Lines • segments: lists of edge points representing straight or curved lines • pixeltype: isolated, starting, interior, ending, junction, corner DC & CV Lab. CSIE NTU

  16. 11.3 Linking One-Pixel-Wide Edges or Lines DC & CV Lab. CSIE NTU

  17. 11.4 Edge and Line Linking Using Directional Information • edge linking: labeled pixels with similar enough directions form chains • edge linking chains identified as arc segment with good fit to curvelike line DC & CV Lab. CSIE NTU

  18. Take a Break DC & CV Lab. CSIE NTU

  19. 11.5 Segmentation of Arcs into Simple Segments • extracted digital arc: sequence of row-column pairs 4-neighbor or 8-neighbor • arc segmentation: partitions extracted digital arc to fit straight/ curved line • endpoints of subsequences: corner points or dominant points DC & CV Lab. CSIE NTU

  20. 11.5 Segmentation of Arcs into Simple Segments • basis for partitioning process identication of all locations with • sufficiently high curvature (high change in tangent angle to length) • enclosed by subsequences fitting different straight lines or curves DC & CV Lab. CSIE NTU

  21. 11.5 Segmentation of Arcs into Simple Segments • simple arc segment: straight-line or curved-arc segment • techniques: from iterative endpoint fitting, splitting to using tangent angle deflection, prominence, or high curvature as basis of segmentation DC & CV Lab. CSIE NTU

  22. 11.5.1Iterative Endpoint Fit and Split • one distance threshold d* • pixel with farthest distance to line AC is B and distance ≥d* then split DC & CV Lab. CSIE NTU

  23. 11.5.1Iterative Endpoint Fit and Split DC & CV Lab. CSIE NTU

  24. 11.5.2Tangential Angle Deflection • another approach: identify locations where two line segments meet • exterior angle between two line segments: change in angular orientation DC & CV Lab. CSIE NTU

  25. 11.5.2Tangential Angle Deflection DC & CV Lab. CSIE NTU

  26. 11.5.2Tangential Angle Deflection • caution: spatial quantization at small distances can completely mask direction DC & CV Lab. CSIE NTU

  27. 11.5.3Uniform Bounded-Error Approximation • segment arc sequence into maximal pieces whose points deviate ≤ given amount • optimal algorithms: excessive computational complexity DC & CV Lab. CSIE NTU

  28. 11.5.3Uniform Bounded-Error Approximation DC & CV Lab. CSIE NTU

  29. 11.5.4Breakpoint Optimization • after initial segmentation: shift breakpoints to produce better segmentation • first: shift odd final point i.e. even beginning point • then: shift even final point i.e. odd beginning point DC & CV Lab. CSIE NTU

  30. 11.5.5Split and Merge • first: split arc into segments with error sufficiently small • then: merge successive segments if resulting segment sufficiently small error • third: try to adjust breakpoints to obtain better segmentation • repeat: until all three steps produce no further change DC & CV Lab. CSIE NTU

  31. 11.5.6Isodata Segmentation • iterative isodata line-fit clustering procedure: determines line-fit parameter • then each point assigned to cluster whose line fit closest to the point DC & CV Lab. CSIE NTU

  32. 11.5.7Curvature • geometry of circular arc segment defined by and 􀀀 DC & CV Lab. CSIE NTU

  33. 11.5.7Curvature DC & CV Lab. CSIE NTU

  34. 11.5.7Curvature • : curve represented parametrically, • : arc length going from to DC & CV Lab. CSIE NTU

  35. 11.5.7Curvature DC & CV Lab. CSIE NTU

  36. 11.5.7Curvature • : unit length tangent vector at measured clockwise from column axis DC & CV Lab. CSIE NTU

  37. 11.5.7Curvature • : unit normal vector at DC & CV Lab. CSIE NTU

  38. 11.5.7Curvature • : curvature defined at point of arc length s along curve • : change of arc length • : change in tangent angle DC & CV Lab. CSIE NTU

  39. 11.5.7Curvature • natural curve breaks: curvature maxima and minima • curvature passes: through zero local shape changes from convex to concave • surface elliptic: when limb in line drawing is convex • surface hyperbolic: when its limb is concave DC & CV Lab. CSIE NTU

  40. 11.5.7Curvature • surface parabolic: wherever curvature of limb zero • cusp singularities of projection: occur only within hyperbolic surface DC & CV Lab. CSIE NTU

  41. 11.5.7Curvature • Nalwa, A Guided Tour of Computer Vision, Fig. 4.14 􀀀 DC & CV Lab. CSIE NTU

  42. 11.6 Hough Transform • Hough transform: method for detecting straight lines and curves on images • Hough transform: template matching DC & CV Lab. CSIE NTU

  43. 11.6.1Hough Transform Technique • The Hough transform algorithm requires an accumulator array whose dimension corresponds to the number of unknown parameters in the equation of the family of curves being sought. DC & CV Lab. CSIE NTU

  44. Finding Straight-Line Segments • : for straight lines does not work for vertical lines • : slope, : intercept • : where and used • : perpendicular distance from the line to the origin • : the angle the perpendicular makes with the x-axis DC & CV Lab. CSIE NTU

  45. Finding Straight-Line Segments DC & CV Lab. CSIE NTU

  46. Finding Straight-Line Segments • use , and , to determine if point falls into cell an accumulator array quantized in this fashion DC & CV Lab. CSIE NTU

  47. Finding Straight-Line Segments DC & CV Lab. CSIE NTU

  48. Finding Straight-Line Segments • Fig. 7.18(a) • Fig. 7.18(b) • Fig. 7.18(c) • Fig. 7.18(d) • Gonzalez and Woods Digital Image Processing Fig. 7.18 􀀀 • Fig. 7.19(a) • Fig. 7.19(b) • Fig. 7.19(c) • Fig. 7.19(d) Gonzalez and Woods Digital Image Processing Fig. 7.19 􀀀 DC & CV Lab. CSIE NTU

  49. Finding Straight-Line Segments • Fig. 7.18(a) • Fig. 7.18(b) • Fig. 7.18(c) • Fig. 7.18(d) • Gonzalez and Woods Digital Image Processing Fig. 7.18 􀀀 • Fig. 7.19(a) • Fig. 7.19(b) • Fig. 7.19(c) • Fig. 7.19(d) Gonzalez and Woods Digital Image Processing Fig. 7.19 􀀀 DC & CV Lab. CSIE NTU

  50. Finding Circles • : row • : column • : row-coordinate of the center • : column-coordinate of the center • : radius • : implicit equation for a circle DC & CV Lab. CSIE NTU

More Related