From (6.4.7), as k approaches approaches , which is the result of inverse filtering provided that

	(6.4.8)

To the extent that which satisfies (6.4.8) can be found, (6.4.5) can be used in performing inverse filtering. One advantage of the iterative procedure is that it can be stopped after a finite number of iterations. The result after a finite number of iterations is not in general the same as that of inverse filtering, but it is less sensitive to the presence of noise in some cases.

Figure 6.12 illustrates the performance of inverse filtering. Figure 6.12(a) shows an original image of 512 x 512 pixels. Figure 6.12(b) shows the original image blurred by a Gaussian-shaped point spread function. The resulting image size is larger than 512 x 512 pixels, but is windowed by a 512 x 512-point rectangular window. The model of the degraded image in the case is

	(6.4.9)

Figure 6.12(c) shows the image processed by inverse filtering. The processed image is computed by

	(6.4.10)

where is the DFT of and is obtained by

	(6.4.11)

The DFT and inverse DFT sizes used are 512 x 512. In the absence of noise and very small , inverse filtering works quite well despite the fact that in (6.4.9) is affected by windowing.