På svenska

Presentation Information     2011-05-20 (13:15)   •  The seminar room at Vi2

Speaker Feng Sitao
Type Master thesis presentation
Title Image Analysis on Wood Fiber Cross-Section Images
Abstract Lignification of the wood fibers has a significant impact on wood property. So as to do a quantitative analysis of the areas of the lumen (L), normally lignified cell wall (NL) and the highly lignified cell wall (HL) of the compression wood cell, a segmentation method should be developed to identify the three regions of each wood cell in wood fiber cross-section images. In this thesis, the segmentation results of these wood cells are provided by a crisp segmentation method. In order to refine the given crisp segmentation results, two fuzzy segmentation methods: Iterative Relative Multi Objects Fuzzy Connectedness (FC) and Weighted Distance Transform on Curved Space (WDT) are applied to segment the regions L, NL and HL of each compression wood cell. The crisp results are used for multi-seeds selection. An evaluation of the three automatic segmentation methods would be implemented and the manual segmentation results are serving as the ground truth to evaluate the performance of the three automatic methods.