Generalized Hough Transform

Slides:



Advertisements
Similar presentations
A Multi-Resolution Technique for Comparing Images Using the Hausdorff Distance Daniel P. Huttenlocher and William J. Rucklidge Nov Presented by.
Advertisements

Wavelets Fast Multiresolution Image Querying Jacobs et.al. SIGGRAPH95.
1 Hough transform Some Fourier basics: –Nyquist frequency: 1/2 , with  the difference between time samples. If signal is bandwidth limited below Nyquist.
1 Model Fitting Hao Jiang Computer Science Department Oct 6, 2009.
Reminder Fourier Basis: t  [0,1] nZnZ Fourier Series: Fourier Coefficient:
CS 376b Introduction to Computer Vision 04 / 16 / 2008 Instructor: Michael Eckmann.
Image Enhancement in the Frequency Domain Part I Image Enhancement in the Frequency Domain Part I Dr. Samir H. Abdul-Jauwad Electrical Engineering Department.
Some Properties of the 2-D Fourier Transform Translation Distributivity and Scaling Rotation Periodicity and Conjugate Symmetry Separability Convolution.
Fitting a Model to Data Reading: 15.1,
Image Features, Hough Transform Image Pyramid CSE399b, Spring 06 Computer Vision Lecture 10
Camera parameters Extrinisic parameters define location and orientation of camera reference frame with respect to world frame Intrinsic parameters define.
Transforms: Basis to Basis Normal Basis Hadamard Basis Basis functions Method to find coefficients (“Transform”) Inverse Transform.
כמה מהתעשייה? מבנה הקורס השתנה Computer vision.
CSE 185 Introduction to Computer Vision
Computer Vision - Fitting and Alignment
National Center for Supercomputing Applications University of Illinois at Urbana-Champaign Image Features Kenton McHenry, Ph.D. Research Scientist.
Alignment and Matching
: Chapter 14: The Frequency Domain 1 Montri Karnjanadecha ac.th/~montri Image Processing.
Spectral Analysis AOE March 2011 Lowe 1. Announcements Lectures on both Monday, March 28 th, and Wednesday, March 30 th. – Fracture Testing –
Digital Image Processing Homework II Fast Fourier Transform 2012/03/28 Chih-Hung Lu ( 呂志宏 ) Visual Communications Laboratory Department of Communication.
1 Chapter 5 Image Transforms. 2 Image Processing for Pattern Recognition Feature Extraction Acquisition Preprocessing Classification Post Processing Scaling.
Under Supervision of Dr. Kamel A. Arram Eng. Lamiaa Said Wed
Shape Analysis and Retrieval Structural Shape Descriptors Notes courtesy of Funk et al., SIGGRAPH 2004.
Axial Flip Invariance and Fast Exhaustive Searching with Wavelets Matthew Bolitho.
Signal Processing and Representation Theory Lecture 4.
Generalized Hough Transform
Signal Processing and Representation Theory Lecture 3.
Fitting: The Hough transform
Geometry Vocabulary- transformation- the change in the size, shape, or position of a figure. translation- slide--every point in a figure moves the same.
Chapter 5: Neighborhood Processing
Inverse DFT. Frequency to time domain Sometimes calculations are easier in the frequency domain then later convert the results back to the time domain.
Edge Detection and Geometric Primitive Extraction Jinxiang Chai.
ELE 488 Fall 2006 Image Processing and Transmission ( )
Fourier and Wavelet Transformations Michael J. Watts
Partial Shape Matching. Outline: Motivation Sum of Squared Distances.
$100 $200 $300 $400 $500 $100 $200 $300 $400 $500 $100 $200 $300 $400 $500 $100 $200 $300 $400 $500 $100 $200 $300 $400 $500 $100 $200 $300.
Digital Image Processing Lecture 8: Image Enhancement in Frequency Domain II Naveed Ejaz.
Spatial Filtering (Chapter 3) CS474/674 - Prof. Bebis.
Jean Baptiste Joseph Fourier
Linear Filters and Edges Chapters 7 and 8
FOURIER THEORY: KEY CONCEPTS IN 2D & 3D
University of Ioannina
Linear Filters and Edges Chapters 7 and 8
Line Fitting James Hayes.
Sample CT Image.
Fitting: The Hough transform
Lecture 7: Image alignment
Fourier and Wavelet Transformations
Recognition using Nearest Neighbor (or kNN)
A special case of calibration
All about convolution.
Image Processing, Leture #12
Spatiograms Versus Histograms for Region-Based Tracking
Objective Describe how changing slope and y-intercept affect the graph of a linear function. Translations, Rotations, and Reflections. What will happen.
Image Processing, Leture #14
Brief Review of Recognition + Context
An Integrated Approach to Protein-Protein Docking
7.2 Even and Odd Fourier Transforms phase of signal frequencies
CSSE463: Image Recognition Day 26
CSSE463: Image Recognition Day 26
Hough Transform.
CSSE463: Image Recognition Day 27
CSSE463: Image Recognition Day 26
Edge Detection Today’s readings Cipolla and Gee Watt,
Finding Basic Shapes Hough Transforms
Integrated Math One Module 7 Test Review.
Midterm Exam Closed book, notes, computer Similar to test 1 in format:
CSSE463: Image Recognition Day 26
CS5760: Computer Vision Lecture 9: RANSAC Noah Snavely
The Frequency Domain Any wave shape can be approximated by a sum of periodic (such as sine and cosine) functions. a--amplitude of waveform f-- frequency.
Presentation transcript:

Generalized Hough Transform

Correlation as a base of Generalized Hough Transform In order to match a part of a model to a whole, we can use correlation to find the optimal aligning transformation:

Calculating Correlation For each translation, compute the correlation between the target and the translated query:

Calculating Correlation For each translation, compute the correlation between the target and the translated query:

Calculating Correlation For each translation, compute the correlation between the target and the translated query:

Calculating Correlation For each translation, compute the correlation between the target and the translated query:

Calculating Correlation For each translation, compute the correlation between the target and the translated query:

Best match in Correlation For each translation, compute the correlation between the target and the translated query: Best match

Accumulator Space for Correlation Accumulator space like in Hough For each translation, compute the correlation between the target and the translated query:

Complexity of calculating Correlation Complexity for binary n * n grids with O(n) non-zero points: Brute Force: O(n4) – for each of O(n2) translations, compute the O(n2) dot product.

Complexity of calculating Correlation Complexity for binary nxn grids with O(n) non-zero points: Brute Force: O(n4) – for each of O(n2) translations, compute the O(n2) dot product. Fast Integration: O(n3) – for each of O(n2) translations, compute the O(n) dot product.

Complexity of calculating Correlation Complexity for binary nxn grids with O(n) non-zero points: Brute Force: O(n4) – for each of O(n2) translations, compute the O(n2) dot product. Fast Integration: O(n3) – for each of O(n2) translations, compute the O(n) dot product. Fourier: O(n2 logn) – compute the FFT, multiply frequency components, compute the IFFT.

Correlation as Voting The value of the correlation at the point (x0,y0) is: This can be understood in two ways: g(x-x0, y-y0) is the function g(x, y) translated by (x0,y0) g(x-x0, y-y0) is the function g(x0, y0) translated by (x,y) and flipped about the origin

Correlation as Voting In the second approach, we treat the pixels as bins and for every possible translation (x,y), we do the following: Translate g by (x,y) Flip g Scale by the value f(x,y) Update the values of all the bins by the values of the transformed g.

Generalized Hough Transform Correlation as Voting

Complexity of Correlation Complexity for binary n * n grids with O(n) non-zero points: Brute Force: O(n4) – for each of O(n2) translations, compute the O(n2) dot product. Fast Integration: O(n3) – for each of O(n2) translations, compute the O(n) dot product. Fourier: O(n2 logn) – compute the FFT, multiply frequency components, compute the IFFT. Fast Voting: O(n2) – for each of O(n) points on the boundary, cast O(n) votes.

The main idea of a Generalized Hough Transform When we compute the correlation by voting, we spend most of the time casting bad votes. Use extra shape information (e.g. gradients) to cast fewer votes: O(n) complexity: For each of O(n) points on the boundary, cast O(1) votes.