Combining of Extraction Butterfly Image using Color, Texture and Form Features
Dhian Satria Yudha Kartika, Darlis Herumurti, Basuki Rahmat, Anny Yuniarti, Hendra Maulana, Fetty Tri Anggraeny · 2020
Combining the extraction results in digital image processing is an innovation in research. Previous research stated that the extraction was carried out separately for each feature, including color, shape, and pattern. Previous research for color feature extraction was used for koi fish, for shape feature extraction in flowers, and pattern feature extraction on batik. In this study, feature extraction results were combined using a dataset of insects, namely butterflies. The number of datasets used was 890 types of butterflies, which were divided into 10 classes. Before feature extraction and merging, the normalization process is carried out with the aim that the data used is clean from noise. The color feature extraction process uses the RGB and HSV methods by maintaining the color quality of the objects owned by faster computation and color quantization. The extraction of shape features using the LBPROT (local binary pattern rotational excess) method allows the dataset to be in multiple positions to obtain maximum results. The extraction of pattern features with region props allows the region of each object to be obtained. All of the feature extraction results are then tested to get an accuracy value of 75%, while the precision, recall, and f-measure are 76.4 for precision and 75 for recall, and 75.5 for f-measure. This research will continue to be tested with different methods and data.