Light source estimating device, light source estimating method, and imaging device and image processing method

Abstract

A light source estimating method for estimating the color characteristics of a light source with an unknown photographing scene from a sensor response in order to improve a color reproduction quality such as a white balance adjustment, wherein, in a projection conversion unit (6) using parameters obtained by computation that can be colorimetrically approximated from the known spectral sensitivity characteristics of an imaging means (4) with respect to a sensor response value (5) and from the assumed spectral characteristics of a test light source (1), a scene is projected on an image distribution (9) in an evaluation space independent of a photographing light source (2), and an evaluation unit (10) is used to estimate a correct photographing light source (2) by rating a reasonability for a plurality of test light sources (1) based on the distribution condition of the sample value of the projected scene.

Description

technical field [0001] The present invention relates to a light source estimation device, a light source estimation method, an image pickup device, and an image processing method in which, for example, an image pickup device having a plurality of different spectral sensitivity characteristics uses a sensor response obtained when an image of an unspecified arbitrary object is taken value to estimate the spectral characteristics of the color representing the unknown photographic light source illuminating the subject. Background technique [0002] The light incident on the human eye to enable vision is due to a portion of the radiant energy of the illumination, which is reflected by the object being seen and propagated through the air; although the human visual system cannot directly measure the properties of the object and the illumination, it does so even in Objects can also be identified with a certain degree of reliability under lighting with unknown color properties. This...

AI Technical Summary

Problems solved by technology

[0011] There are also studies in which the spectral characteristics of the light source and the object surface are approximated by a linear model of few dimensions by assuming that the object surface is a reflection model of a diffuse surface, using a vector space different from that of the sensor response, by linear calculation to try to reconstruct (non-patent reference 4), and there are studies in which constraints such as the spectral reflectance of the object surface must be physically in the range 0 to 1 are imposed to select a light source with a high probability ( Non-Patent Reference 5); however, in generalized image pickup systems with few response channels, these systems do not independently provide sufficient estimation performance

Images