Mass Function Initialization Rules for Ganoderma Infection Detection by Tomography Sensor

M.S. Mazliham (Malaysia), P. Loonis (France), and A.S. Idris (Malaysia)

Keywords

DempsterShafer Mass Initialization,Ganoderma infection detection, Tomography Image Analysis,Mass distribution.

Abstract

The problem of the detection of Ganoderma fungus infec tion inside oil palm tree is addressed with the objective to develop a method to help facilitate decision-making pro cess by the biologists. More specifically, this study focuses on the initialization of mass functions to be applied in the combination rules using Dempster Shafer theory. In prac tice, a tomography image of the cross section of the palm under study is used to detect abnormalities in the palm stem. Using expert knowledge, the abnormalities are di vided into three categories which conditioned the overall palm health : intact, decay or degradation caused by Gan oderma infection and decay or degradation not caused by Ganoderma. These categories are considered as singletons of the frame of discernment θ. A comparison between sev eral method to distribute basic probability assignment (bpa) or mass to the possible elements of the power set of the frame of discernment based both on the feature extracted from the tomography and the expert knowledge for each zone is discussed. Finally, the combination method be tween masses obtained for the different zones to generate the distributed belief map is then proposed.

Important Links:



Go Back