Dispersion state of fibers in fiber reinforced composites is one of the important factors dominating mechanical properties of them. In this study, we attempt to evaluate the dispersion state via an image analysis. In order to express the dispersion state the degree of dispersion (D(n)) was employed. For the calculation of D(n), a sample image which both sides were 2N pixel is prepared and divided by grids into segments. Herein, grid size can be 2i (i=3,4,…,N-1) pixel. After thresholding, the target area is extracted from the sample image and the area ratio in each segment is derived. The degree of dispersion is defined as coefficient of variation between the area ratios. This method was applied for two different cross-section images of carbon fiber (CF) reinforced composites (CFRP). One of them showed a marblelike structure and it was more inhomogeneous than another one. Derived D(n) reflects dispersion state of fibers in the sample images. D(n) of the marblelike sample image was higher than that of more homogenous one with all grid size. Moreover, relationship between D(n) and n showed size of aggregates in the sample images. Therefore D(n) will be a useful parameter to evaluate the dispersion state by the comparing sample images to a reference image.
