SoBinaryCorrelationProcessing2d Class Reference
[Pattern Recognition]

SoBinaryCorrelationProcessing2d engine More...

#include <ImageViz/Engines/GeometryAndMatching/PatternRecognition/SoBinaryCorrelationProcessing2d.h>

Inheritance diagram for SoBinaryCorrelationProcessing2d:

Classes
class	SbCorrelationDetail
	Results details of image correlation. More...
Public Types
enum	OffsetMode { OFFSET_1 = 0, OFFSET_2 = 1, OFFSET_4 = 2, OFFSET_8 = 3 }
Public Member Functions
	SoBinaryCorrelationProcessing2d ()
Public Attributes
SoSFImageDataAdapter	inBinaryImage
SoSFImageDataAdapter	inBinaryKernel
SoSFEnum	offsetMode
SoImageVizEngineOutput < SoSFImageDataAdapter, SoImageDataAdapter * >	outMatchingImage
SoImageVizEngineAnalysisOutput < SbCorrelationDetail >	outResult

Detailed Description

SoBinaryCorrelationProcessing2d engine

SoBinaryCorrelationProcessing2d performs the logical correlation between a binary image and a binary kernel.

Notations: $a(and)b \leftrightarrow a \wedge b$ $a(or)b \leftrightarrow a \vee b$ $(not)a \leftrightarrow \overline{a}$

The binary correlation between image I and kernel K is defined as:

$O1(n,m)=\sum_{i=1}^{kx} \sum_{j=1}^{ky} \left( K(i,j)\wedge I(n+i-\frac{kx}{2},m+j-\frac{ky}{2}) \right)$

$O2(n,m)=\sum_{i=1}^{kx} \sum_{j=1}^{ky} \overline{\left( K(i,j)\vee I(n+i-\frac{kx}{2},m+j-\frac{ky}{2}) \right)}$

$O(n,m)=O1(n,m)+O2(n,m)$

At the end of the processus, the correlation image is normalized between -1 and 1.

Figure 1: Example of SoBinaryCorrelationProcessing2d

When a part of the kernel lies beyond the edge of the image, the correlation is not computed on the border. In which case, the values are set to -3e38.

Note: The SoBinaryCorrelationProcessing2d filter does not accept a mask AOI attached to the kernel image.