Identification of aromatic coconuts using image processing and machine learning techniques

The paper develops an efficient and accurate method for detecting fresh aromatic coconuts. Coconuts have a nearly cosmopolitan distribution due to human action in using them for agriculture. At present, the only way to determine whether a coconut is aromatic or not is by tasting it. By implementing the IAC (Identification of Aromatic Coconuts) method as proposed in this research, it is possible to identify the aromacy through non-invasive mechanisms with the help of image-processing techniques. The brightness of the image has to be adjusted accordingly for actual implementation. The underlying principle is that the color of the region of interest at the bottom part of the coconut shell is correlated to its age. Segmentation is done on the image via K-Means. The region of interest in RGB color is converted in to HSV and the Threshold is applied to it. After that the amount of white pixels in each layer on the image are measured using Polynomial Regression to obtain the predicted value of aromacy.