2015 · 32 citations · 9 references
ComputingComputer ScienceComputer VisionFeature EngineeringMachine LearningFeature ExtractionImage ClassificationImage Feature EngineeringImage TextureOptical Character RecognitionPattern RecognitionAccuracyBackpropagationColor Feature ExtractionColor FeaturesGray-level Co-occurrence MatricesImage Feature ExtractionStatistical Image FeaturesVisual Feature Extraction
Indonesian's Batik is one of culture heritage that recognized around the world. Batik has many variations of motif based on their region. This paper discusses feature extraction methods for classifying batik motifs in digital images. A single feature extraction method may result feature vector that is similar for two different images. In this research, the using of Gray Level Co-occurence Matrix (GLCM) and statistical color RGB features can represent more characteristics in extracting batik images. The extracted features vectors are furthermore classified into motifs using Backpropagation Neural Network with several scenarios for testing the level of accuracy. Some experiment by using single feature and combination of GLCM and statistical color RGB features show that the best result for classifying batik image is the combination of feature extraction with rate of precision 90.66%, recall 94% and accuracy 94%.
9
Statistical and structural approaches to texture
R.M. Haralick · Proceedings of the IEEE · 1979 · 5.7K citations
How many hidden layers and nodes?
Demetris Stathakis · International Journal of Remote Sensing · 2009 · 441 citations