Fog removing device and image generating method

Patent metadata

Field	Value
Publication number	US-10102614-B2
Application number	US-201615371228-A
Country	US
Kind code	B2
Filing date	Dec 7, 2016
Priority date	Jun 12, 2014
Publication date	Oct 16, 2018
Grant date	Oct 16, 2018

Abstract

To allow fog removal even in a densely foggy image. A fog density calculating unit 11 calculates the fog density of an input image given, by using a separated illumination light component. A reflectance component fog removing unit 13 performs fog removal on a reflectance component calculated by a reflectance calculating unit. An illumination light component fog removing unit 14 performs fog removal on the separated illumination light component. Here, the degree of fog removal by the reflectance component fog removing unit 13 is higher than the degree of fog removal by the illumination light component fog removing unit 14. Thereby, the degree of fog removal of the reflectance component can be raised without significantly raising the level of fog removal of the illumination light component.

First claim

What is claimed is:

1. An image generating method comprising:
separating, in a fog-containing image, a reflectance component and an illumination light component;
removing fog based on a fog density separately determined for the reflectance component;
removing fog based on the fog density for the illumination light component; and
synthesizing an image from which fog has been removed, by using a reflectance component after the fog removal and an illumination light component after the fog removal, wherein
under a condition that there is not a reflectance component in airglow and that an atmospheric model holds true to the illumination light component, the atmospheric model being expressed by an equation of I(x)=J(x)t(x)+A(I−t(x)) in which I denotes an observation image including fog and is defined by the product of an illumination light component IL and a reflectance component IR in a Retinex theory in which I, J and A satisfy the Retinex theory, J denotes a processed image not including fog, t denotes a fog density, x denotes the coordinate of a target pixel in the observation image and A denotes the airglow,
fog removal is performed separately on the reflectance component and the illumination light component, by applying the Retinex theory to the I, J and A, as JR=(IR*IL−(1−t)AL)/(IL−(1−t)AL) for the reflectance component and as JL=(IL−AL)/t+AL for the illumination light component, wherein JR is a reflectance light component of a processed image not including fog, AL is the airglow due to the illumination light component, and JL is an illumination light component of the processed image not including fog, and
degrees of fog removal are different between the reflectance component and the illumination light component due to different values being used for fog densities or due to different weighting of the reflectance component after fog removal and the illumination light component after fog removal.

2. The image generating method according to claim 1 , wherein the degree of fog removal in the reflectance component is higher than the degree of fog removal in the illumination light component.

3. The image generating method according to claim 1 , wherein the degree of fog removal in the illumination light component is higher than the degree of fog removal in the reflectance component.

4. The image generating method according to claim 1 , wherein the degree of fog removal is due to different values being used for fog densities in respective reflectance component and illumination light component.

5. The image generating method according to claim 1 , wherein the degree of fog removal is due to weighting in respective fog removing process results of the reflectance component and the illumination light component.

Assignees

• Eizo Corp

Inventors

• Nakamae Takashi

Classifications

• G06T2207/10024

  Color image · CPC title

• G06T2207/20192

  Edge enhancement; Edge preservation · CPC title

• G06T5/20

  using local operators · CPC title

• G06T2207/20021

  Dividing image into blocks, subimages or windows · CPC title

• G06T2207/10004

  Still image; Photographic image · CPC title