Development of an algorithm to detect the Bloom effect on the blueberries (#1375)

Authors

LAFITTE CHONG, ADRIANA CRISTINA

GOMEZ AVALOS, NAYELLY MASSIEL

LEÓN LEÓN, RYAN ABRAHAM

Abstract

This report focused on the recognition of blueberries using the segmentation method and the subsequent selection of blueberries identified as good. The image change method from RGB to HSV was applied, which allowed better discrimination of the blueberries in relation to the background and other elements present in the image. Successful segmentation was achieved by creating masks based on Hue, Saturation and Value (HSV) values. Grading based solely on the absolute size of blueberries may not be sufficient to ensure accurate and consistent grading. With the Hough transformation method, blueberries can be better discriminated by approximating them in a circular manner. However, an image filter with acceptable Bloom concentration ranges was used for greater accuracy. In conclusion, this report highlights the success in recognizing blueberries using the segmentation method and presenting the selected blueberries as good. With an average effectiveness of 98%, favorable results can be obtained so that the process mentioned can be used, as valid for the stated objectives.