Moisture Content Determination of Corn Using Red-Green-Blue to Hue-Saturation-Value Color Space with Artificial Neural Network

This paper is focused on non-destructive moisture content determination of a yellow shelled corn. The hardware of the device is composed of Raspberry Pi with Pi Camera and NeoPixel Led Strip. Red-Green-Blue to Hue-Saturation-Value color space conversion was used to get the enhanced image of the shelled corn. After the color space conversion, feature extraction was applied. Thus, it produced a statistical data to use as an input parameter in the artificial neural network. In the creation of the model, multi-layer feedforward neural network with backpropagation was used in determining the moisture content of corn. A total of 30 samples were tested into the device. It was concluded that the moisture content values measured by the device are comparable to the values measured by the conventional method which is by using a moisture content tester for grains, with maximum percent difference of only 4.35%.