A Stable Algorithm of Box Fractal Dimension and Its Application in Pore Structure

Box counting method (BCM) is often used for image fractal dimension (FD) estimation as it is applicable for images with or without self-similarity. However, BCM may be affected by noise, the size of image and original box size, etc. In particular, when the image size can't be exactly divided by the box size, it makes the conventional BCM instable and the characterization of pores inaccurate. The paper presented a modified BCM and the stability of several BCMs was discussed. An adaptive threshold was used for box counting in the proposed method, which effectively reduced the impact of noise and image size. The experimental results demonstrate that the modified BCM is more accurate and reliable compared with the conventional ones. It can be well applied for the characterization of the pore structure of porous materials. The parameter complexity level (CL) gained from combining area-FD with curve-FD can characterize the irregular degree of pores better.