X-ray Image Processing Of Log Defets Based On The Minimum Box-counting Fractal Dimension

Our country is facing so many problems, such as the lake of forest resources, the huge popularity and the low average forest resources share. The issue how to use the limited forest resources reasonably, filling the need of huge popularity, requires researchers to increase forest resources utilization rate. The wood nondestructive testing is put forward to solve this problem, which accurately distinguish the log defect and its size without destroying its structure.There are many methods of nondestructive log testing. In this paper, the X-ray detection was chosen as a mean for wood nondestructive testing, based on the comparison of the methods. Also, the paper introduced the working principle of X-ray detection, and its experiment condition, procedure and effect, etc. In the process of testing, the log was scanned by X-ray respectively, obtaining its X-ray transmission digital image. The X-ray image was equalized and filtered to show the features of log defect. The defective log was selected for analysis and research. This paper mainly studied two extensive log defects:knot and rot.Fractal theory is widely used in digital image processing. The grey value change of irregular and fractured in image can be well described. Based on fractal theory, the grayscale space of log image can be mapped into a fractal parameter space, which shows the defect edge more clearly and accurately. In this paper, the minimum box-counting dimension method was used to calculate fractal dimension. The log image was divided into small windows, and the fractal dimension of every small window was calculated by minimum box-counting dimension. The set of these fractal dimensions was fractal dimension image of the log. Research shows, the fractal dimension of normal log image part is2, while that of defect and defect edge are less then2. That is, the fractal dimension occurs singularity. The collection of singularities is the log defect.Lots of experiments were carried on in this paper, and the result demonstrates the accuracy and effectiveness of the method that using fractal dimension theory to quantitatively describe the size and shape of log defects. The traditional edge detectors, such as Sobel, LoG and Canny, etc. are more effective in fractal dimension space than in grey value space. This paper provides a new and convenient fractal dimension computational method for log defect image.