Macro-sized basidiomycota species recognition using shape and color features

Large number of species dan lack of knowledge of Indonesian people about macro-sized Basidiomycota (toadstool) cause identification of those mushrooms become so difficult, so that encouragement to take advantage of the useful mushrooms is still very lacking. The purpose of this research is to propose a method to detect the species of mushrooms that is found around people so that they recognize its species directly. The proposed method is the combination of shape and color feature. Centroid-Contour Distance is used to extract shape feature and Color Moment is used to extract color feature. These two features are combined to recognize the mushroom. The accuracy was tested by using 25% of training data as testing data to get the ROC Curve. In this experiment, the area under curve is known as the quality of the accuracy that shows the result 0 to 1. The result showed 0,65751 accuracy on 20 species of mushroom with 5 data each species, and 0,69649 accuracy on 32 species. Then the result on 32 species showed 0,69649 accuracy with 5 data each species, 0,65407 with 7 data each species, and 0,60008 accuracy with 10 data each species.