| dc.description.abstract | In recent time, the ability to identify and classify leaves becomes a great need for taxonomist to know the diversity of plants (Hickey et al 1999). Identification can be done by identifying features of morphology and texture of the leaves or also with a combination of both. Annisa (2009) implemented the feature extraction approach to obtain the basic characteristic morphological image derived from the leaf blade and the co-occurrence matrix for texture feature extraction. Morphological features that obtained were diameter, leaf area, leaf perimeter, smoothing factor, form factor and perimeter ratio of diameter. Texture features were obtained energy, inverse difference moment, entropy, maximum probability, contrast, correlation, and homogeneity. This study used morphological features approach to obtain morphological features and co-occurrence matrix for texture features. Therefore, the evaluation of image retrieval was done by using more complete morphological features with additional morphological features of physiological length and physiological width. Both morphological features are very useful to help identify the characteristics of a leaf blade. This research approach has been successfully implemented morphological features. Value of image retrieval evaluation leaves increased with the addition of two morphological characters. In this research, with additional morphological features and physiological length and physiological width resulted in morphological traits 0.2083, the texture feature values 0.1864 and character of Bayesian Network 0.2055. So that the average value of precision increased by 0.0137 for morphological features and by 0.0069 for combined value of morphological and textural characteristics (Bayesian Network). | en |
