Skip to Main content Skip to Navigation
Theses

Etude de l'organisation spatiale du tissu conjonctif par analyse d'images basée sur une approche multiéchelles. Application à la prédiction de la tendreté de la viande bovine

Abstract : The aim of this work is to characterize muscular tissue by evaluating its quality from imaging data. More precisely, we intend to develop prediction tools of bovine meat tenderness based on arti cial vision process, by studing intramuscular connective tissue, which contributes in a signi cant manner to the intrinsic toughness of meat. The images of muscle slices were made using two types of lighting: polarized white light and ultraviolet. Our contribution to analyze the meat images is based on a multiscale approach. Two methods of segmentation were proposed, both based on discrete wavelet transform, in particular the "à trous" algorithm. The rst one is based on the universal thresholding and the second, on the k-means algorithm applied on the image resulting from the summation of wavelet planes. Another section of this work deals with parameters extraction and decision. Information retained is the distribution of considered object sizes of the meat connective network. Statistical tools which are the linear regression and neural networks were applied on data issued from stages of image processing. The nal model retained for tenderness prediction was determined according to a maximization criterion of R2. The choice of the number of parameters was based on a cross validation criterion (Leave one out). Prediction results, derived from data base of the study, are very encouraging highlighting an undoubtedly correlation between the images parameters and sensory quality of meat particulary tenderness.
Document type :
Theses
Complete list of metadatas

Cited literature [105 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00728810
Contributor : Camille Meyer <>
Submitted on : Thursday, September 6, 2012 - 4:39:50 PM
Last modification on : Thursday, February 25, 2021 - 10:34:02 AM
Long-term archiving on: : Friday, December 16, 2016 - 10:39:18 AM

Identifiers

  • HAL Id : tel-00728810, version 1

Citation

Mohammed El Jabri. Etude de l'organisation spatiale du tissu conjonctif par analyse d'images basée sur une approche multiéchelles. Application à la prédiction de la tendreté de la viande bovine. Analyse fonctionnelle [math.FA]. Université Blaise Pascal - Clermont-Ferrand II, 2008. Français. ⟨NNT : 2008CLF21831⟩. ⟨tel-00728810⟩

Share

Metrics

Record views

330

Files downloads

1176