Research of Image Correction Algorithm Using the Structures of Face Image (I: Verification of Benefit about Remarkable Point)

Abstract

We have analyzed an exquisite technique of image retouch by detecting remarkable points of expert photographer. These remarkable points indicate an attention point of observed image, and will change the point according to the retouch task. It is very important information to explore the expert skill and analyze the operation procedure. Therefore, in this paper, we constructed the system to measure the remarkable point accurately, and investigated the relationship between remarkable points and retouch operation by comparing expert and amateur. As the result, it was clear that expert subjects observed all over the image and watched the unnatural feature such as bruise, sagging skin, and blood vessel. On the other hand, with the retouch task for color adjustment, it became clear that the expert referred to the surrounding white point for adjustment of skin color, as opposed to the amateur almost observing only the skin color domain. Therefore, we conclude that the measurement of remarkable points is very useful for detecting the attention points and analyzing retouch of color reproduction with referred points.