Quality Characterization of Tissue and Newsprint Paper based on Image Measurements; Possibilities of On-line Imaging

The paper industry is at a turning point worldwide. The consumption of printed products is assumed to decrease significantly because of electronic gadgets and the internet. Furthermore, the importance of forests and woods in the global warming issue is significant and causes limitations to papermakers to produce cheap products. However at the same time the end users demand high quality products at reasonable prices. Therefore, the cost-efficiency in papermaking process will play a more significant role in the future. This thesis proposes a set of image analyses and methods which can be implemented in present on-line imaging systems, without a massive effort. The imaging systems can measure the paper accurately and in a versatile manner which improves the control of the paper machine and thus enhances the cost efficiency of the papermaking process. The creping process which generates the wavy microstructure to tissue paper - a creping pattern - mainly determines the quality properties of the tissue paper. This thesis applied the photometric stereo method with 2D spectral analysis to characterize the versatile properties of creping pattern. The creping process also affects the free fiber ends and pinholes in tissue paper. The free fiber end detection method was based on the number of shadows of fibers. The shadows were made visible by removing the creping pattern from the images of tissue paper with photometric stereo method. The pinhole detection method was based on the linearly polarized light transmitted through paper and the camera with rotating polarizer. It is anticipated based on the thesis that the characteristics of creping patterns and the number of pinholes are possible to be measured on-line by using existing imaging technology in paper machines. The web inspection systems installed on paper machines measure the web-wide light transmittance of paper. At present they detect and classify defects such as holes and dirt particles and estimate some simple paper characteristics such as light transmittance and formation. This thesis proposed methods to utilize the on-line transmittance images more versatile in newsprint grade. The cross directional shrinkage profile and the small scale structural characteristics such as the light transmittance, formation, length scale of formation and orientation parameters of paper were estimated based on the statistical measures and the features of frequency distribution in the 2D spectrum. The uncertainty of basis weight estimate approximated from the transmittance images was studied with correlation analysis and the Kubelka-Munk equations. It was noticed that the initial quality of the on-line images was insufficient for evaluating formation characteristics and basis weight reliably. However capturing a complete set of calibration and reference images will also enable the formation and basis weight estimation of paper and thus increasing the value of web inspection system significantly