Research On Wood Defect Recognition By Analyzing Tomography With Fractal Method

As a fast wood assessing technique,non-destructive wood testing has attracted broad attention.Wood tomography with advantages of efficiency,non-destruction and intuition,has been widely used in manufacturing and trades.However,current research mostly focuses on utilizing the properties of wave propagation to produce accurate tomographic images,leaving the image and its analysis insufficiently studied.Therefore in practice,analysis of these images has not been automated,but still relies on human judgment.In this paper,the feature of wood tomographic images is studied,proposing a defect recognition method based on fractal theory.This method uses the Differential Box-Counting(DBC)method to analyze the pre-processed wood tomographic images.For the defected images,image segmentation techniques are used to extract the defected region and to provide relevant parameters such as the size of the defects.The main contributions of this paper are as follows:(1)A fractal-based automatic wood defect recognition method is proposed.Color space conversion is used to reduce the disruption the non-defective features in the images,based on the properties of wood tomographic images.In addition,DBC is conducted to partition the image into blocks,and the dimension number of each block is computed.In the end,a defected region is found by testing whether the dimension number lies within a specific interval.(2)The method to compute the properties of a wood defect,such as size,is introduced,realizing quantified recognition.For a defected wood image,equalization and filtering methods are first used to reduce the noise and to enhance the details.And then,two dimensional Otsu method is performed to segment the defective region.As a result,the size of the defect is obtained from cumulative histogram.(3)Our method is verified in a small sample set.The number of dimension of a good wood region ranges from 2.0000 to 2.1792,while for a defective region it lies between 2.0083 and 2.3867.In a boundary region,this number ranges between 1.6024-1.9794.The result of computed wood decay using our method is very close to the rate which is computed manually.Hence,the experiment result shows high accuracy of the computed wood decay method as presented in this article,providing a new method in detecting wood defects automatically.