Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.sao.ru/drabek/CCDP/TUTORIAL/Imqualiti/tutorial-quality-acutance.htm
Дата изменения: Wed Nov 11 22:51:02 1998
Дата индексирования: Tue Oct 2 13:42:57 2012
Кодировка:
Image quality-acutance

Tutorial-Head1.GIF (19571 bytes)


Acutance

The concept of acutance was introduced in the 1940's following the realization that resolving power does not necessarily correlate with image quality on objects in general. Acutance is an objective measure of the physical characteristics that give the subjective impression of sharpness in an image. It incorporates the intensity distribution across the image of a knife-edge, and includes the effects of any non-linearities. As was shown in Figure 1, the knife-edge ideal image would have a rectangular profile suggested by the broken line; any real image has a gradually sloping profile determined by the edge-spread function. It has been shown that neither the average slope nor the maximum slope correlates well with subjective image impression.

Crane proposes a more useful measure of accutance. To measure acutance, first the extent of the edge is found by determining two point E1 and E2 where the gradient is very low. Regions far away from the geometrical edge image are exclude so that the regions of very low gradient do not weight the results. The x distance from A to B is then divided into n small intervals; DX1 and DX2 etc and the corresponding values of intensity are noted. The gradients DI/DX are the squared, and the average gradient is found as

The acutance is given as

SMT acutance system modulation transfer acutance has introduced by Crane. The SMT takes into account that the sharpness is related to the square of the standard deviation of a Gaussian Spread function. Crane used the square of the area under the MTF curve to accurately predict subject image quality [].

Image Edge Profile (IEP) Acutance can be measured by determining the edge profile at the normal to an edge boundary.

where m is the mean derivative of the jth boundary pixel and f(i), b(i), i 1,2,3,4 .. are the foregraound and the background pixels along the normal. After the derivatives are calculated the rms. gradient is computed on the boundary. This value is then normalized by the maximum possible rms. derivative. The expression for the IEP acutance is given by