Classify the grid pattern in an original image.
More...
#include <GridClassifier.h>
|
| GridClassifier () |
| Single constructor.
|
|
void | classify (bool isGnuplot, const QPixmap &originalPixmap, const Transformation &transformation, int &countX, double &startX, double &stepX, int &countY, double &startY, double &stepY) |
| Classify the specified image, and return the most probably x and y grid settings.
|
|
Classify the grid pattern in an original image.
This class uses the following tricks for faster performance:
- FFT is used for "fast correlations" in frequency space rather than graph space
- FFT initialization/shutdown housekeeping is done once
- Rather than a combinatorial search of grid line start, step and count, we exploit the periodicity of the FFT to search start and step as the first step, and then as a separate second step we search count. In the first step, the periodicity means the repeating grid lines wrap around the end of the end of the image back around to the start of the image - so the grid line count is not even relevant. In other words, the searches are START X STEP + COUNT rather than START X STEP X COUNT
Definition at line 26 of file GridClassifier.h.
The documentation for this class was generated from the following files: