Regresar

Classification of Andean Chocho (Lupinus Mutabilis Sweet) by Shape and Color Using Artificial Vision

Abstract:

In Ecuador, the classification of the cooked grain of Lupinus Mutabilis Sweet, Andean lupine, according to its color and shape is done manually. This work shows an alternative classification using artificial vision. The system is developed on Matlab’s Simulink visual programming platform, running on a Raspberry Pi B + board, incorporated into a test bench that maintains constant lighting during image acquisition. The image processing is carried out under the HSV color model, which allows to intuitively choose the defined ranges of its three components of hue, saturation and value to discriminate areas in the image based on their color. To determine the acceptance by color of the grain, the average value of the components is compared with thresholds that are calculated experimentally. The grains are evaluated by shape based on their eccentricity. The classification system indicates suitable grains according to their shape and color with an accuracy of 81.83% compared to those selected by human operators.