1 / 91

By: Navid Einackchi

به نام خدا. Isfahan University of Technology Electrical and Computer Department. Master Thesis of Computer Engineering- Artificial Intelligence and Robotic. Image alignment and stitching using Object Recognition Methods. By: Navid Einackchi. Supervisors: Dr Rasoul Amirfattahi

hall
Download Presentation

By: Navid Einackchi

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. به نام خدا Isfahan University of Technology Electrical and Computer Department Master Thesis of Computer Engineering- Artificial Intelligence and Robotic Image alignment and stitching using Object Recognition Methods By: Navid Einackchi Supervisors: Dr Rasoul Amirfattahi Dr javad Askari Advisor: Dr M. Saraee Spring 2007

  2. Presentation Process • Introduction • Definition • Aligning concepts • Stitching Concepts • Object Recognition Problem • Salient Feature • What is Descriptor • Matching • Proposed Method for Aligning • Using Harris Detector • Using SIFT Detector • Stitching • Concluding and Suggestions Some slides are from other sources

  3. Presentation Process • Introduction • Definition • Aligning Concepts • Stitching Concepts • Object Recognition Problem • Salient Feature • What is Descriptor • Matching • Proposed Method for Aligning • Using Harris Detector • Using SIFT Detector • Stitching • Concluding and Suggestions

  4. Presentation Process • Introduction • Definition • Aligning Concepts • Stitching Concepts • Object Recognition Problem • Salient Feature • What is Descriptor • Matching • Proposed Method for Aligning • Using Harris Detector • Using SIFT Detector • Stitching • Concluding and Suggestions

  5. تصوير حاصل Image Alignment and Stitching • Definition • Applications • Mosaic Image • Panorama • Virtual Environment • Robotic

  6. Image Alignment and Stitching • Definition • Applications • Mosaic Image • Panorama • Virtual Environment • Robotic

  7. Image Alignment and Stitching • Definition • Applications • Mosaic Image • Panorama • Virtual Environment • Robotic

  8. Presentation Process • Introduction • Definition • Aligning Concepts • Stitching Concepts • Object Recognition Problem • Salient Feature • What is Descriptor • Matching • Proposed Method for Aligning • Using Harris Detector • Using SIFT Detector • Stitching • Concluding and Suggestions

  9. Aligning Concepts • Image Transformations • Applicable Transformations • Find transformations • How many parameters? • Calculating Parameters

  10. Image Transformations • Transition • Number of Parameters: 2 • Number of Points: 1

  11. Image Transformations • Euclidean • Number of Parameters: 3 • Number of Points: 2

  12. Image Transformations • Similarity • Number of Parameters: 4 • Number of Points: 2

  13. Image Transformations • Affine • Number of Parameters: 6 • Number of Points: 3

  14. Image Transformations • Perspective • Number of Parameters: 8 • Number of Points: 4

  15. Image Transformation Computation • Direct • Mapping one image into another using different parameters • Define an Error Function based on pixel intensity difference • Using Corresponding points • Obtaining Corresponding points • Computing Transformation parameters using corresponding points

  16. Direct Method Error Function

  17. Direct Method Error Function ؟

  18. Direct Method

  19. Using Corresponding points

  20. Using Corresponding points

  21. Using Corresponding points

  22. Using Corresponding points • How many points? • How to select? • Using human • Automatically

  23. Presentation Process • Introduction • Definition • Aligning Concepts • Stitching Concepts • Object Recognition Problem • Salient Feature • What is Descriptor • Matching • Proposed Method for Aligning • Using Harris Detector • Using SIFT Detector • Stitching • Concluding and Suggestions

  24. Stitching • Stitching • Choices • Final Plane • Pixel weighting

  25. Presentation Process • Introduction • Definition • Aligning Concepts • Stitching Concepts • Object Recognition Problem • Salient Feature • What is Descriptor • Matching • Proposed Method for Aligning • Using Harris Detector • Using SIFT Detector • Stitching • Concluding and Suggestions

  26. Object Recognition Methods • Local Feature Detection • Edge • Corner • Hole • Interest Regions Matching • Appearance Matching • Geometric Matching • Finding Object Relations in Image

  27. Presentation Process • Introduction • Definition • Aligning Concepts • Stitching Concepts • Object Recognition Problem • Salient Feature • What is Descriptor • Matching • Proposed Method for Aligning • Using Harris Detector • Using SIFT Detector • Stitching • Concluding and Suggestions

  28. Salient Features • Features of Feature! • Invariants against transformations • Ability to explain the image • Ability to being Matched (Repeatability) • Selected Features • Corners • Holes

  29. Harris Detector Plain No changes in every direction Edge High changes in Edge Direction Corner High changes in every direction

  30. Harris Detector(Continued) Fast change Direction (max)-1/2 Slow change direction (min)-1/2

  31. Harris Detector(Continued) Edge2 >> 1 2 “Corner”1,2both big1 ~ 2 Edge1 >> 2 Plain 1

  32. Harris Detector(Continued) • Harris measurement

  33. Harris Detector(Continued) • Harris measurement response 2 “Edge” R < 0 “Corner” R > 0 Local Maximum bigger than Threshold “Plain” “Edge” |R| small R < 0 1

  34. Threshold R R x x Harris Detector(Continued) • Features • Invariant against rotation • Invariant against intensity

  35. Harris Detector(Continued) • Example

  36. Scale Invariant • Scale Problem

  37. Scale Invariant Finding Appropriate Window Size

  38. - = Scale Selection • Using Differential of Gaussian Filters • Laplacian of Gaussian (LoG) • Difference of Gaussian (DoG) Window thst maximize filter

  39. Automatic Scale Selection

  40. Automatic Scale Selection

  41. مقیاس  DoG  y x  DoG  SIFT Detector • Bulb Detector • Invariant against scale • Using DoG • Bulb detection • Scale detection

  42. Bulb position Scale position SIFT Detector • Detecting Bulb and Scale Simultaneously • Bulb position = maximum in plain • Scale = maximum in third dimension

  43. Presentation Process • Introduction • Definition • Aligning Concepts • Stitching Concepts • Object Recognition Problem • Salient Feature • What is Descriptor • Matching • Proposed Method for Aligning • Using Harris Detector • Using SIFT Detector • Stitching • Concluding and Suggestions

  44. Descriptor • Vector which describes region of interest point • Pixels • Histogram • Differential • Feature of Descriptor • Invariant against image distortion (view, angle, intensity) • Able to express similarities and differences

  45. 2p 0 SIFT Descriptor • Based on value and direction of gradients • 128 dimensional Vector • Construction • Calculation of value and direction of gradient of each pixel • Making direction histogram of gradients • Rotating interest region based on dominant direction • Dividing the region into 16 region • Constructing direction histogram for each region

  46. Construction Example

  47. DescriptorFeatures • Invariant against • Rotation • Position • Affine changes in intensity(I  Ia+ b)

  48. Presentation Process • Introduction • Definition • Aligning Concepts • Stitching Concepts • Object Recognition Problem • Salient Feature • What is Descriptor • Matching • Proposed Method for Aligning • Using Harris Detector • Using SIFT Detector • Stitching • Concluding and Suggestions

More Related