PITH ESTIMATION ON ROUGH LOG END IMAGES USING LOCAL FOURIER SPECTRUM ANALYSIS

The location of the pith is an important feature of cross sections from tree logs. Images of log ends can be taken at little cost and at almost every stage in the log processing chain. Analysing images of rough log ends automatically requires robust pith estimation. This work evaluates two pith estimation algorithms using four different local Fourier Spectrum analysis methods. Proving that size, selection and amount of annual ring sections have an impact on both algorithms, this work contributes to existing literature. In comparing experiments for pith estimation for digital images of rough log ends and computer tomography (ct) cross section images, this paper highlights the difficulties for pith estimation on rough log end images. Finally, our results show that peak analysis and principal component analysis for local Fourier Spectrum analysis achieve the best accuracy and timing performance for pith estimation on rough log end images.