back

APPLICATIONS IN AGRO-ALIMENTARY INDUSTRIES AND RESEARCH

Image Processing

 

TwinBro

El Programa Estatal para Impulsar la Investigación Científico-Técnica y su Transferencia y el Subprograma Estatal de Transferencia de Conocimiento, dentro del Plan Estatal de Investigación Científica, Técnica y de Innovación para el período 2021-2023, en el marco del Plan de Recuperación, Transformación y Resiliencia, ha publicado la concesión de ayudas públicas a proyectos de colaboración público-privada, destinadas a financiar proyectos de desarrollo experimental en colaboración entre empresas y organismos de investigación.
Estas ayudas se enmarcan en la inversión I3, denominada «Nuevos proyectos I+D+I Publico Privados, Interdisciplinares, Pruebas de concepto y concesión de ayudas consecuencia de convocatorias competitivas internacionales, I+D de vanguardia orientada a retos de la sociedad, Compra pública pre-comercial», del componente 17 «Reforma institucional y fortalecimiento de las capacidades del Sistema Nacional de Ciencia, Tecnología e Innovación» del Plan de Recuperación, Transformación y Resiliencia de España y contribuirán al cumplimiento de los objetivos asociados a la misma, de acuerdo con la normativa reguladora del Mecanismo de Recuperación y Resiliencia y del Plan de Recuperación a nivel europeo y nacional.
TWINBRO es un proyecto que tiene como objetivo global el diseño, desarrollo e integración de una arquitectura Gemelo Digital, aplicable a la cría de pollos de engorde, compuesta por una granja física en el espacio real, una representación digital de esa granja en el espacio virtual y la sincronización entre el espacio virtual y el real para transferir datos e información mediante tecnologías IoT. El Gemelo Digital estará basado en una arquitectura de información habilitadora compuesta de tecnologías de inteligencia artificial, aprendizaje automático, captura y transformación eficiente de datos masivos, simulación y sistemas de apoyo a la toma de decisiones.
El proyecto está liderado por TRIPLEALPHA INNOVATION SL., empresa tecnológica especializada en analítica de datos aplicada al ámbito de Producción y Operaciones, y cuenta con la colaboración de la empresa IMASDE AGROALIMENTARIA SL, líder en I+D para la industria agroalimentaria, y con el Grupo de Investigación LIA2: Laboratorio de Inteligencia Artificial Aplicada de la Universidad de Vigo.

 

FAETÓN

The principal objective of this research is to investigate whether there is a correlation between the features of the images taken by an electron microscope, the production conditions of the bio-polymers, and their macroscopic properties.

nanotubes

Example of an image taken with a electron microscope that shows a flake of bio-polymers or nanotubes. We processs such an image through different steps in order to equalize its background, to reduce its noice, to binarize the gray scales into black&white, and eventually to compute its skeleton.

thinned image

 Example of a skeleton image of a bio-polymer. Such an image is further processed in order to find its connected components and to calculate certain features such as length or width distributions.

Example of the width distribution of the nanotubes of the main component in the image as shown above.

As a side effect within this research, we have developed a general improvement for iterative and parallelizable thinning algorithms. We call the new method guided thinning and the skeletons obtained with the modification are much better centralized within the shapes. The results that we achieve when the method is incorporated into standard iterative thinning algorithms are shown in this image gallery. The corresponding research article is currently under peer review.

 

El grupo LIA2 participó en el Cluster de Investigación e Transferencia Agroalimentario do Campus Auga (CITACA).

Hyperspectral image processing for quality control in the potato industry

The overall objective of this project was the design and analysis of image processing and hyperspectral imaging techniques to automate certain tasks related to quality control in the potato industry which are developed manually until now. The study covered the classical stages of computer vision, but now applied to hyperspectral images: design and implementation of an image acquisition system, preprocessing and segmentation through various image processing techniques, feature extraction, and classification.

Some of the tasks that have been developed are:

  • Classification of tubers depending on their external defects (greening, rotten, etc.).
  • Detecting hollow heart in potatoes using hyperspectral imaging.
  • Estimating area affected by common scab in potatoes using hyperspectral imaging.

We provide to the research community the database (.rar-file 1.52 GiB, md5-checksum, and Readme-file) of a set of hyperspectral images which we used to detect the hollow heart and other deseases in potatoes using hyperspectral information in the infrared spectrum (900 nm to 1700 nm).

Classification and counting of pollen grains

Palynological data are used in a wide range of applications such as paleoclimatic reconstruction, allergenic process, and food technology. Fossil pollen spectra are used for paleoenvironmental reconstruction in the Quaternary and in the geological past. The count of pollen in the air allows to suggest safety treatments to patients who are allergic to pollen. Pollinic analysis of honeybees is used to determine their geographical origins. The principal tasks in almost all of these applications are classification and counting, which are laborious and time consuming due to the fact that they are done manually by highly skilled experts. A high level of training is needed to obtain accurate identification results. The morphology of the grains is fundamental to identify the type of pollen. The key parameters are shape, polarity, symmetry, aperture, size, exine stratification, and sculpturing type of the grain.

We are interested in discriminating species of the same family Urticaceae, namely, Parietaria Judaica, Urtica Urens, and Urtica Membranacea which is not achieved in routine pollinic analysis. Our attempt is to classify and count the grains of pollen in a slice that represents the visual information reported by an optical microscope. The method consists of two steps: detection of pollen grains in the slice based on the Hough transform, and their classification. The detection step includes two phases: finding the location of the grains and extracting their silhouette. The classification step also includes the phase of computing the feature vector that represents each pollen grain which includes area, diameter, compactness, roundness, holes, thickness, elongatedness, eccentricity, several statistical measures and fourier descriptors. For classification itself we use minimum distance classifiers, multilayer perceptron, and support vector machine.

The final result obtained was 86% of correct pollen classification using the most convenient combinations of features and classification algorithms. The largest values of correctness has been reached for the Urtica Membranacea with 98%, but the value fell down to 82% for the Urtica Urens and to 76% for the Parietaria Judaica.

Automatic classification of honey bee pollen

Automatic classification of wine leaves


Financed Projects


Publications


People Involved


Student Work


Press Articles


Resources


Partners

More about IMASDE
IMASDE
Ir a web
More about TripleAlpha
TripleAlpha
Ir a web
More about Agencia Estatal de Innovación
Agencia Estatal de Innovación
Ir a web