Wood Quality and Geometry Measurements Based on Cross Section Images

Abstract

This thesis discusses the application of the image analysis methods for the cross section surface images of wood pieces, such as disc and block samples or logs and boards. The analysis methods have been developed in the laboratory environment for the measurement of wood geometry and quality properties. The focus has been in annual ring analyses, and the algorithms were based on texture analysis methods in order to obtain high robustness. Main algorithms are automatic. The texture based methods work quite well in varying imaging conditions, with chain-saw cut surfaces and moderate defects. In addition, a few geometry and color based algorithms were developed, but they are more sensitive to lighting variations, defects and roughness of wood cross section surface, when automated. The thesis discusses wood properties, the present wood measurement techniques and image based measurement application. Based on the review, technology gaps that can be covered with image based measurements are identified. The required calibration and implementation of algorithms, such as local annual ring width and orientation measurement, counting annual rings, pith locating and defect detection are described. Uncertainties affecting the measurement methods are analyzed. The purpose of the thesis is to create the basis for the development of wood cross section image based measurements for industrial applications.