Blame view

3rdparty/boost_1_81_0/libs/gil/example/harris.md 1.02 KB
977ed18d   Hu Chunming   提交三方库
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
  # Harris Corner Detection
  
  Harris corner detection capabilities in GIL are demonstrated by the program `harris`, compiled from the sources `example/harris.cpp` and `hvstack.hpp`.
  
  ## Synopsis
  
  `harris input.png window-size discriminant harris-threshold output.png`
  
  - The first parameter must be the full path to an existing image in the PNG format for `harris` to process
  - The second parameter is the size of the window containing the pixels to analyse and must be an odd number (e.g. 9 for a 3x3 matrix)
  - The third parameter is the empirically-defined discriminant constant, usually between 0.04 and 0.06
  - The fourth parameter is the harris threshold used to identify the optimal values
  - The fifth and last parameter is the full path to the output image of `harris`. The directory will *not* be created and must exist.
  
  ## Specific requirements
  
  ### Build requirements
  
  - A C++ compiler compliant with C++14 or above
  - The PNG library installed and configured
  
  ### Execution requirements
  
  - `harris` has no specific execution requirements.