Font Size: a A A

The Research On Families And Genera Of Trees Computer Identification Based On Parameters Of Annual Ring

Posted on:2011-11-07Degree:MasterType:Thesis
Country:ChinaCandidate:Q ZhuFull Text:PDF
GTID:2178360308471322Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
Annual ring is one of the most crucial natural attributes of wood, Traditional annual ring analysis method brings low efficiency, processing method tedious,time-consuming and hard sledding, and the stochastic undulation influence for the measuring technique. With the leap development of computer processing technology, new technology promote automation and intelligence in Traditional annual ring measuring technique and wood recognition.The annual ring image segmentation was put forward based on watershed algorithm, this dissertation puts forward a new method of adaptive region growing algorithm and background removal in order to deal with the regional fault and background noise,so as to classify the families and genera of trees.Some parameters of images were extracted based on preprocessed image, and it made the foundation of families and genera of trees recognition.This dissertation proposed the methods and the thoughts which are suitable for the content of this subject. It's mainly study content as follows:Preprocessing method of annual ring image based on Watershed Algorithm is advanced in this paper. It is much suitable to solution Over-Segmentation by using CIE1976 L*u*v* uniform color space and region merging.By using this method,the basic shape features of original image are preserved and clear edge of annual ring is detected. on that basis, a new background removal algorithm was put forward.Thus favorable precondition is created for the features extraction of annual ring.A new adaptive region growing algorithm in color annual ring image segmentation was put forward,which makes good use of color and spatial information of regions near the seeds, and combined with the edge information so as to adaptive changes growth conditions,solve the problem of the regional fault. Compared to the traditional algorithms, the new one combined with watershed algorithm can select the seeds automatically, the complete shape features of original image are preserved and clear edge of annual ring is detected. This method has very good noise immunity.Thus favorable precondition is created for the features extraction of families and genera of trees.This dissertation analyzes the image of the annual ring, and distills the related parameter, such as the earlywood percent, latewood percent, mean sensitivity and so on, builds a database of families and genera of trees,and completes the identification of families and genera of trees base on discrimination of minimal difference of parameters.According to the parameters of annual ring physical meanings and image features,families and genera of trees computer identification based on parameters of annual ring system was designed, which provided the theory basis for further analyzing some questions about physics mechanical properties of families and genera of trees.
Keywords/Search Tags:Watershed, Region growing, L~*u~*v~* color space, Mean sensitivity
PDF Full Text Request
Related items