Skip to Main content Skip to Navigation

Analyse automatique d'images de populations microbiennes

Abstract : A common goal of biotechnological research and of commercial production is the definition of optimum conditions for achieving predetermined objectives. This goal is usually translated into the problem of finding the optimum control strategy that will produce the desired end-product. To achieve optimum performance an on-line supervision of growth is essential. In this works we present the methods which allow to analyze images of microbial populations in order to identify the cells. This identification is based on the edges obtained by active contours, using a specific image energy. The initialisation and the snakes convergence into the curvature of the budding cell are difficult and were possible by a meticulous analysis of the image energy. To differentiate the cells as single or budding, some methods were compared, including fuzzy clustering, wavelet transform and the change of sign of the radius of the curvature. Consequently, the approximation of the edge of each cell by an ellipsoidal model, was carried out by comparing the results of the genetic algorithms, the hough transform and least squares. The proposed supervision system of microbial population was implemented in a new software of image analysis which is a new tool for description of the cells.
Document type :
Complete list of metadatas

Cited literature [77 references]  Display  Hide  Download
Contributor : Anne Bergez <>
Submitted on : Tuesday, January 10, 2006 - 11:39:21 AM
Last modification on : Friday, January 10, 2020 - 9:08:09 PM
Long-term archiving on: : Friday, September 14, 2012 - 5:05:35 PM


  • HAL Id : tel-00011336, version 1


Laurent Manyri. Analyse automatique d'images de populations microbiennes. Interface homme-machine [cs.HC]. INSA de Toulouse, 2005. Français. ⟨tel-00011336⟩



Record views


Files downloads