Résumé
The analysis of wood anatomical sections is of great interest for understanding the growth anddevelopment of plants. We propose a novel method for automatically identifying and characterizing radialfiles in wood microscopic images of gymnosperms. A key point is to be able to assign a priori reliabilitycoefficient to the results, particularly for statistical processing in large-scale analyses. We describe in thispaper the principle used to establish reliability coefficients to evaluate the radial file identification processand the geometrical measurements of cells and their components.