Abstract

Image analysis of handmade paper is a cross-field research of traditional process, history of Chinese paper and information technology.The objective of the research is to, by applying digital image processing, image analysis technology to the study of handmade paper, realize nodamage or minimaldamage measurement, improve automation and intelligence of analysis of handmade paper, and provide a new method for preservation of paper relics.

Methods used in this research include literature reviewing, inspection and practices, experts interviewing, microscopy experiments and computer analysis.Through research and experiments, plenty materials such as historical data of paper-making, identification of papermaking materials, microscopic images of handmade paper, process of making traditional handmade paper, methods of scientific analysis of pager, are obtained.These materials provide the basis for the research and based on these, three research projects were carried out:study on the correlation between the age and the material of a paper, nodamage measurement of evenness of handmade paper, measurement of ingredients based on color features of images.

Study on the correlation between the age and the material of paper is based on the history of Chinese papermaking and the microscopic image analysis knowledge, and summarizes the correlation between types of raw materials, fiber characteristic and paper age.Based on the development of papermaking raw materials, the development of the raw materials of Chinese papermaker is summarized in chronological order; then, by virtue of microscopic images and fiber characteristics of typical raw materials of paper of different era, the correlation between fiber characteristics, raw materials and paper age is established.The new method has been used and explored in authentication of paper relics and has good applicability and reliability.

Nodamage measurement of evenness of handmade paper uses microscopic transmission images of handmade paper.Based on Fourier transform theory, a statistical model of fiber distribution and paper evenness evaluation parameters are designed.Simulation experiment and paper experiment prove that the method is effective and reliable.Fourier transform analysis is a method of nodamage measurement, which has some significance and application value in the field of paper relics.

Raw materials of Chinese traditional papermaking are plant fibers and mixing many types of fibers is very common.Types and percentage of raw material determine the performance of the paper.Traditional component analysis relies on manual observation, which is time-consuming and laborious.Identification of raw materials generally relies on microscopic observation.The samples need be dyed before the microscopic observation, which makes different types of fibers shows different colors.Using fiber color differences, this thesis designed a K-means color clustering model. Images are divided into disjoint K clusters based on the color characteristics. Models were applied to measure the color fiber image and grayscale metal image.Experiments showed that when the color differences are significant, K-means clustering algorithm was consistent with the actual results for both color image and gray scale image.For images color difference between which is significant, image analysis method based on color characteristics provides a new choice for component measurement, which is more simple and efficient way.

Key Words:Handmade paper Image analysis Paper age analysis Evenness analysis Composition measurement