Fig. 1. The neighborhood of an average filter; the point ‘’ is the selected pixel. (a) The neighborhood is constituted of four pixels; this is the simplest mode. (b) The neighborhood is constituted of eight pixels; the precision will be better than with four pixels.

Fig. 2. The neighborhood of a median filter; the center point is the selected pixel. (a) The neighborhood is constituted of nine pixels; this is the most commonly used mode. (b) The neighborhood is constituted of 25 pixels; the precision will be better than for the mode with nine pixels but requires more processing time.

Fig. 3. Original images of the near and far fields. (a), (b) Near-field images. (c), (d) Far-field images.

Fig. 4. Images of threshold processing without a filter. (a), (b) Near-field images. (c), (d) Far-field images.

Fig. 5. Near-field images after multiple filters. (a), (e) The original images captured by near-field CCD. (b), (f) After use of a median–linear filter with 25 pixels. (c), (g) After use of an average filter with eight pixels. (d), (h) The images after threshold processing has been performed.

Fig. 6. Far-field images after multiple filters. (a), (e) The original images captured by far-field CCD. (b), (f) After use of an average filter with eight pixels. (c), (g) After use of a median–linear filter with 25 pixels. (d), (h) The images after threshold processing has been performed.

Fig. 7. The results of the iteration method without a filter.

Fig. 8. The result of the adaptive variable step method with filter denoising.