RUBBER LATEX VESSEL DETECTION FROM MICROSCOPIC CROSS-SECTION IMAGES

Abstract

This thesis presents the methods to detect and segment rubber latex vessels from gray scale microscopic cross-section images using polynomial curve-fitting with maximum and minimum stationary points. Polynomial curve-fitting is used to detect the location of rubber latex vessels from an image of a non-dyed cross-section slice which was taken by a digital camera through microscope lens. The rubber latex vessels are then segmented from an image using maximum and minimum stationary points with morphological closing operation. Two species of rubber trees of age between one to two years old are sampled namely, RRIM600 and RRIT251. Two data sets contain 30 microscopic cross-section images of one-year old rubber tree’s stems from each species are used in the experiments and the results reveal that most of the rubber latex vessel areas can be segmented correctly.