| 摘要 |
There is provided a method in which a human image including a face of a person is extracted from a color original image at a high probability and an amount of exposure is determined so that the human image is reproduced at the most suitable density. Photographing information is read from a negative film, a color original image is divided, on the basis of the photometric data of the three colors of red, green and blue, into small regions in which color and density are the same for each of pixels on the original image, and characteristic values which indicates characteristics of the small regions are calculated. Subsequently, a small region which is estimated, on the basis of photographing information, to be a background whose size is largely different from the size of a main image, and a small region which is determined to have a low probability of being the main image on the basis of positional information are obtained, and these small regions are eliminated as non-main regions. A small region in a predetermined range of color or density corresponding to the human image is extracted, as an effective region, from other small regions than the eliminated non-main regions. The density of the human image is estimated by using the average luminance in the effective region, and the estimated density is outputted.
|
