Image Processing Of Log Defects Based On Fractal Dimension And Fractional Brownian Motion

On the condition of nondestructive appearance and structure of log, testing internal defects of log correctly, it is an effective method to utilize forest resource, and it is significant for selecting log scientifically. X-ray testing method was chose to detect logs inner defects. Log images were obtained according to the intensity differences of the ray through logs. Then we established x-ray image mathematical model by using fractal theory to process and analyze log images, in order to judge whether the logs had defects or not and the details of defects. In this paper two common defects--hollow and rotten knot were studied.Two fractal methods were introduced to detect log internal defects in this paper. One is the box-counting dimension method. Segmented log x-ray images into several sub areas, and then calculated the fractal dimensions of every sub areas one by one. There were differences in fractal dimension between the background regions and defects edges. So when we analyzed x-ray image, fractal dimension could be as reference to distinguish the defections from the normal regions. Extracted singular fractal dimension, their set was the margins of image. The other method is fractional Brownian motion, it is an effective method to describe natural scene. Log image can be considered as the different grayscale fractional Brownian motion image is made from the random motion. If Hurst exponent is bigger than one, then fractal dimension is less than two. It is impossible that fractal dimension is less than its topology dimension, so the Hurst exponent is singular. The areas of singular Hurst exponent are the edges of the log images. Thus we considered Hurst exponent equal to one as the threshold. If Hurst exponent is bigger than one, then the regions are the edges. If it is less than one, the regions are the backgrounds. At last we calculated the fractal dimension, the set of all the singular fractal dimension values was the edges of log defects.Firstly, preprocessed the log x-ray images, the methods included histogram equalization, smooth filtering, edge detection and so on. Features of log image were enhanced and more suitable for later processing. We studied the connotative feature information extracted from x-ray images data based on the fractal theory. Describe irregular degree of log x-ray image in quantity by using the value of fractal dimension. This experiment collected one hundred log sample images. It can be known from the experimental results that these two methods efficiency amount to 95%. They are also suitable for detect other common log defect types. These two methods have important significance for promoting the application of fractal theory. At the same time this study provides a new method for digital image processing and edge detection.