Features and Extractors

Lire supports global features as well as local features. There are proper interfaces for each type of features. Following, you can find a brief description of the available interfaces, that you can follow to implement and integrate your own features.

LireFeature

LireFeature is the basic interface for all content based features and extends the FeatureVector interface. The FeatureVector interface contains the getFeatureVector() method, which you can use in order to get the feature's vector as a double[] array. LireFeature is extended by two interfaces, the GlobalFeature and the LocalFeature.

Global Features

Global features, like CEDD or AutoColorCorrelogram, should implement the GlobalFeature interface. That interface extends both LireFeature and Extractor interfaces. This means that a global features holds both the feature and the extract() method.

Local Features

On the other hand, due to the fact that local feature extractors create a lot of local features, there are two interfaces that should be used in that case. The LocalFeatureExtractor, which extends the Extractor interface and contains the extract() method, and the LocalFeature, which extends the LireFeature interface.