• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1Basics
2------
3
4Here are basic concepts that might help to understand documentation
5written in this folder:
6
7Convolution
8~~~~~~~~~~~
9
10The simplest way to look at this is "tweaking the input so that it would
11look like the shape provided". What exact tweaking is applied depends on
12the kernel.
13
14--------------
15
16Filters, kernels, weights
17~~~~~~~~~~~~~~~~~~~~~~~~~
18
19Those three words usually mean the same thing, unless context is clear
20about a different usage. Simply put, they are matrices, that are used to
21achieve certain effects on the image. Lets consider a simple one, 3 by 3
22Scharr filter
23
24``ScharrX = [1,0,-1][1,0,-1][1,0,-1]``
25
26The filter above, when convolved with a single channel image
27(intensity/luminance strength), will produce a gradient in X
28(horizontal) direction. There is filtering that cannot be done with a
29kernel though, and one good example is median filter (mean is the
30arithmetic mean, whereas median will be the center element of a sorted
31array).
32
33--------------
34
35Derivatives
36~~~~~~~~~~~
37
38A derivative of an image is a gradient in one of two directions: x
39(horizontal) and y (vertical). To compute a derivative, one can use
40Scharr, Sobel and other gradient filters.
41
42--------------
43
44Curvature
45~~~~~~~~~
46
47The word, when used alone, will mean the curvature that would be
48generated if values of an image would be plotted in 3D graph. X and Z
49axises (which form horizontal plane) will correspond to X and Y indices
50of an image, and Y axis will correspond to value at that pixel. By
51little stretch of an imagination, filters (another names are kernels,
52weights) could be considered an image (or any 2D matrix). A mean filter
53would draw a flat plane, whereas Gaussian filter would draw a hill that
54gets sharper depending on it's sigma value.
55