@Reference(type=Inproceedings,
author={"Che-Hua Yeh","Yuan-Chen Ho","Brian A. Barsky","Ming Ouhyoung"},
title="Personalized Photograph Ranking and Selection System",
year="2010",
booktitle="Proceedings of ACM Multimedia",
pages={"211","220"},
month="October",
customData={"location","Florence, Italy"})
public class RuleOfThirds
extends Object
implements org.openimaj.image.analyser.ImageAnalyser<org.openimaj.image.MBFImage>, FeatureVectorProvider<DoubleFV>
I've assumed that the distances to the power-points should be normalized with respect to the image size - this isn't explicit in the paper, but given that the sigma of the gaussian is fixed, it seems likely...
| Constructor and Description |
|---|
RuleOfThirds()
Construct a new
RuleOfThirds with the default settings for the
YehSaliency algorithm. |
RuleOfThirds(float saliencySigma,
float segmenterSigma,
float k,
int minSize)
Construct a new
RuleOfThirds with the given values for the
YehSaliency algorithm. |
| Modifier and Type | Method and Description |
|---|---|
void |
analyseImage(org.openimaj.image.MBFImage image) |
DoubleFV |
getFeatureVector() |
public RuleOfThirds()
RuleOfThirds with the default settings for the
YehSaliency algorithm.public RuleOfThirds(float saliencySigma, float segmenterSigma, float k, int minSize)
RuleOfThirds with the given values for the
YehSaliency algorithm.saliencySigma - smoothing for the AchantaSaliency classsegmenterSigma - smoothing for FelzenszwalbHuttenlocherSegmenter.k - k value for FelzenszwalbHuttenlocherSegmenter.minSize - minimum region size for
FelzenszwalbHuttenlocherSegmenter.public DoubleFV getFeatureVector()
getFeatureVector in interface FeatureVectorProvider<DoubleFV>public void analyseImage(org.openimaj.image.MBFImage image)
analyseImage in interface org.openimaj.image.analyser.ImageAnalyser<org.openimaj.image.MBFImage>