Determining blewah fruit maturity based on skin color texture using feature extraction

Abstract

Blewah (Cucumis Melo L) is the original fruit of Indonesia. Blewah is a similar fruit to Cantaloupe, but difference in skin texture. In Indonesia, many farmers are conducting cultivation of cantaloupe. The period of cultivation of blewah fruits need 50 days after planting period. Purpose of this research is determined blewah maturity based on skin color texture image. Feature extraction process begins with converting RGB image to grayscale image. Feature extraction is processed on the grayscale image with 11 characteristic parameters. 11 characteristic parameter is Mean, Variance, Skewness, Kurtosis, Entropy, Angular Second Moment, Contrast, Correlation, Standard Deviation, Inverse Difference Moment and Homogenity. 11 characteristic parameters are further classified using artificial neural network LVQ. The final weight obtained is used as a reference weight to identify whether the blewah is ripe. Based on the test conducted using 30 data test, the feature extraction method has a good accuracy level. 85,7% accuracy rate obtained by the number of test data as much as 30 images using first ordo feature extraction method. 78,5% accuracy rate obtained by the number of test data as much as 30 images using second ordo feature extraction method