Research Of Wood Cell Image Feature Extraction And Image Databases

Along with the development of the computer technology, the image processing technology, the computer vision and the pattern recognition technology, the computer vision applied to the wood cell recognition domain is the new technology which emerges in recent years. It is day by day widespread in the wood scientific domain. The people has done the massive research to the cell picture division and the automatic diagnosis method, has developed many cells automatic diagnoses and the analysis system. The cell analysis and the automatic diagnosis reduced the subjective disturbance, lightened the correlation personnel's work load, and enhanced the analysis accuracy. Its essential technology is carries on the cell edge outline to withdraw, the edge examination quality directly will interfere with the following analysis work, and its result directly will affect the cell related parameter the analysis.In the view of the situation that certain quantities of cells adhere, overlap and gather when the outline edge of cells is extracted during the research on forming characteristic of wood cells, this paper puts forward the improved two-step inflation model of Snake with which outline of cells is extracted. At the same time, the greedy optimized algorithm is applied to reduce the running time of the procedure. This method finishes the segmentation to the cell cavity of wood at first step, and then to the cell wall of the wood, which enhances the effect of the exterior strength and the precision of capturing the cell outline, and also helps to reduce the impact on the possibility of trapping in the local minimum point of the object energy function.For to resolve the difficult question of extraction computer-aided wood identification based on wood transverse section annual ring micrograph characteristics, it measured the micrograph characteristics by computer vision technology, The main method is described as follows. It determined and analyzed the cellular contour, cellular dimension characteristic, tissue proportion characteristic etc. of the wood transverse section on the basis of the second exploitation of the multicolor image manipulation software. 10 quantitative parameters was gathered and carried on the principal components analysis and other analyses by SPSS software .These findings for the lumber distinction, the tree-classification will offer a new technical means and relevant theory foundation.Because kinds of trees and cells in each annual ring are more, so the size of digital image collections increases rapidly. As the solution, the image database has used. The image database is this kind of system, it can store large quantities of image dates and it's the correlation information in together, and carries on the effective management to them, guaranteed the data the uniformity, the integrity, support each kind of application.