The Vision Package (original) (raw)

Next: Cameras Up: Gandalf: The Fast Computer Previous: Image sequence I/O Contents The vision package includes some higher level image manipulation routines. To be able to use any routine or structure in the vision package use the declaration

  #include <gandalf/vision.h>

but including individual module header files instead will speed up program compilation.

Subsections * Cameras
* Building cameras
* Projecting points and lines
* Adding/removing camera distortion
* Building the camera calibration matrix
* Converting cameras between precisions * Computing the fundamental/essential matrix * Computing a homography between 2D scene and image
* Computing a 2D homography from an array of feature matches
* Computing a 2D affine homography * Computing a homography between 3D scene and image * Smoothing an image using a 1D convolution mask * Smoothing an image using a 2D convolution mask * Feature detection
* Image feature coordinate frames
* Edge detection
* Displaying an edge map
* The Canny edge detector
* Corner detection
* Displaying a corner map
* The Harris corner detector
* Line segment detection
* Displaying a line map
* The Gandalf line detector * Representing 3D rotations
* Quaternion routines
* General rotation routines * Representing 3D Euclidean transformations * Levenberg-Marquardt minimisation
* Robust observations
* Generalised observations
* Levenberg-Marquardt software * Fast Hough Transform
* The Fast Hough Transform (FHT)
* Notation
* The FHT Algorithm
* Example: Line Fitting
* Example: Plane Fitting
* Calculating the Intersection of a Plane and a Sphere
* Calculating the Plane Parameters of a Child Cube
* Formal Statement of the FHT Plane Fitting Algorithm
* Speed Improvement to FHT Line Finder
* Formal Statement of Algorithm
* Comparision of Speed and Memory Requirement with Standard FHT
* Complexity Comparison of FHT and MFHT

2006-03-17