qa.qcri.prepocressor.operations.feat

Class DistributionFeatureExtractor

java.lang.Object

qa.qcri.prepocressor.operations.feat.FeatureExtractor

qa.qcri.prepocressor.operations.feat.DistributionFeatureExtractor

public class DistributionFeatureExtractor
extends FeatureExtractor

This feature extractor implements the distribution features described in "Features for HMM-Based Arabic Handwritten Word Recognition Systems", 2012. For f_1, see RawFeatureExtractor. Note that all feature indices are shifted by 2 (1 (since it starts by 0) + 1 (f_1 is omitted)). WARNING: Assumes constant frame width

Author:

Felix Stahlberg

Constructor Summary

Constructors

Constructor and Description
DistributionFeatureExtractor(Configuration conf)

Method Summary

Methods

Modifier and Type	Method and Description
java.util.List<java.lang.Double>	extractFeatures(org.opencv.core.Mat whole, org.opencv.core.Mat frame) Extract distributional features.

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

DistributionFeatureExtractor

public DistributionFeatureExtractor(Configuration conf)

Method Detail

extractFeatures

public java.util.List<java.lang.Double> extractFeatures(org.opencv.core.Mat whole,
                                               org.opencv.core.Mat frame)

Extract distributional features. Extract features from the given frame. The length of the returned array should be always the same as long as the height of the frame matrix stays the same.

Specified by:

extractFeatures in class FeatureExtractor

Parameters:

whole - whole input frame

frame - current frame from which to extract features

Returns:

list of features or null if an error occurred