Procesamiento de imágenes digitales aplicado al análisis de color y textura del piloncillo de la Huasteca Potosina

Abstract

The present work demonstrates the implementation of image processing techniques to analyze the color and texture of agroindustrial products, in this case, of the piloncillo or panela in its granular and cone variety. A technological tool based on software was developed, capable of quantifying color and texture and thus implementing quality controls to support the commercialization of this product made in Huasteca Potosina towards global markets. An unsupervised classification method (clustering) is proposed to define the color or colors of the piloncillo, called K-means. In the case of texture analysis, second-order statistical methods derived from the co-occurrence matrix and six Haralick textural descriptors are used: contrast, homogeneity, energy, ASM, correlation and dissimilarity. The developed prototype was validated in 24 producing communities of granulated and cone-shaped piloncillo belonging to the municipalities of Tanlajas and Tancanhuitz, S.L.P. The results of the analysis of color and texture have been related to the manufacturing methods that are currently used, to identify improvements that help standardize the production process.

Métodos y proyectos transdigitales

“…La Tabla 1 presenta la caracterización colorimétrica de la corteza de los panqués. iv) Clasificación: para la clasificación de las variables que hacen la mayor aportación a la caracterización del producto, se utilizan herramientas estadísticas como Discriminant Analysis (LDA), k-Nearest Neighbours (k-NN), Decision Trees (DT) entre otros (Lágarra-Altamirano et al, 2019;Vélez et al, 2013).…”

“…iii) Texturales: dimensión fractal, entropía, contraste, segundo momento angular, diferencia del momento inverso, etc. (Lágarra-Altamirano et al, 2019;Arzate et al, 2011). una inversión costosa, debido a que los aditamentos y el software para su funcionalidad también se adquieren por separado.…”