110.

[Cover] [Contents] [Index]

Page 197

Figure 5.6 (a) Build up a column of boxes based on Equation (5.1 1), (b) Intersect the boxes with the image intensity surface to compute n_L(i, j).

The fractal dimension D is the negative of the slope of the least squares line with different value of L and N_L, as shown in Equation (5.8). The algorithm for differential box counting is shown in Figure 5.7.

A third method, known as the triangular prism surface area method, was developed by Clarke (1986) and later used by de Jong and Burrough (1995) as a basis for classification of remotely sensed images. This method is somewhat different from the box-counting approaches described above, but it is based on the same concept of walking dividers. Instead of calculating the number of different scale boxes, the method measures the triangular area needed to cover the image at different scales.

[Cover] [Contents] [Index]