| Summary: | Fruits are one of important agricultural commodities in Indonesia. Banana
is one of fruits commodities with high demand because it has a lot of benefits.
Community needs for domestic and non domestic market on bananas also
followed with guaranteed qualities. With many varieties of banana, Pisang Mas is
sold quite high on retail level compared another bananas variety. Maturity level of
banana is one of the determining factors for quality. Sorting process on Pisang
Mas based on color grade usually depend on human�s perception of color images
composition factor owned by the fruit. The development of the image processing
system, which is combined with a method to apply the artificial neural network
(ANN), enabled for the identification of the level of maturity of Pisang Mas
according to grade more accurately and quickly.
In this study igital image processing refers to two-dimensional images
processing using a computer. ANN is a computational system which the
architecture and operation system inspired by the knowledge about the biological
neuron cells in brain. Pisang Mas are varieties from Kebun Plasma Nutfah Pisang
Yogyakarta. Number of sample is 84 bananas which divided into 2 groups, 56
bananas ( 224 images) as training data and 28 bananas ( 112 images) to testing the
network. Image capturing of each sample conducted on each of four sides. The
parameters used as input to the ANN is mean Red, mean Green, homogeneity and
contrast.
The results in this study show that, with combining image processing and
artificial neural network method, maturity level identification of Pisang Mas
veriety based on USDA standard (green, light green, yellowish green, greenish
yellow, yellow with green tips, yellow and yellow flecked with brown) can be
successfully done. ANN architecture consisted of four cells
|
|---|