Corn quality identification using image processing with k-nearest neighbor classifier based on color and texture features

Abstract

Corn is food crop commodity that is widely used, including as raw material for animal feed. Determination of corn quality at the farm level is often associated with drying time. This method has weaknesses, namely low efficiency, objectivity and level of consistency and also can lead to conflicts between traders and farmers. This study aims to identify the quality of corn using digital image processing based on color and texture features. This research uses Pertiwi-3 and Pertiwi-6 corn varieties. The corn quality identification system uses 7 features input (hue, saturation, value, contrast, correlation, energy, homogeneity) and KNN algorithm as classifiers. The number of image data used are 500 images with a test ratio of 70: 30. This research is able to classify the quality of corn into 10 quality categories. The highest accuracy is obtained at 90.00% when the k value (the nearest neighbor) is 5 and the distance calculation method is Cityblock.