Table 1.

The basic flow of image segmentation and bias field estimation.

Step 1. Set the number of cluster c, the exponent of fuzziness m, stopping condition of the iteration ε > 0, the radius s of local neighborhood window, and the radius u of search window in equation (11).

Step 2. Set the iteration initial value t=0 and initialize bias field b0=1 and the center of cluster V1={v11, v21,…, vc1}.

Step 3. Calculate nonlocal weight coefficient wklt in equation (8) and then obtain the nonlocal weighted value ykt by equation (7).

Step 4. Update the membership degree uikt by equation (14).

Step 5. Update the center of clustering vit by equation (14).

Step 6. Calculate the estimated bias field bkt by equation (15).

Step 7. If satisfying max‖Vt+1 − Vt‖ < ε, then terminate iteration; otherwise, go to step 3 and set t=t+1.