where a is a lower cutoff level and b is a higher cutoff level of a band-pass filter (Fig. 7). Usually, only the high-pass filter is used to calculate the intensity of orientation (b = N).

The influence of the high-pass filtering on orientation intensity can be visualized by the reconstruction of the images using an inverse FFT after the masking of the low-frequency component (Fig. 8).

In the image of the trabecular bone, low-frequency components constitute bulk density of the trabeculae. High-frequency

FIGURE 6 An image of collagen fibers represented in 256 gray levels and its frequency transform (A). Coordinate system in the original image and frequency transform (B).

FIGURE 7 Coordinate transform from Cartesian to polar using a linear interpolation (A). Angular distribution analysis of frequency transform (B). The intensity of orientation at an angle of 6 is calculated as a summation of the power spectrum in a segment in a narrow fan-shaped area (shown as a gray area).

Power spec Dtini

Linear Interpolation

Power spec Dtini

FIGURE 8 Filtering effects on the image of connective tissue. Images are reconstructed by an inverse Fourier transform using frequencies from a to b.

Fine structure such as collagen fiber is more suitable for Fourier analysis.

0 0

Post a comment