Illumination color prediction method, image processing method, apparatus and storage medium
A method for performing illumination color prediction on an image in a neural network model, comprising: inputting an image to the neural network model; extracting a semantic-based illumination color feature of the image and a statistical rule-based illumination color feature of the image; and predicting an illumination color of the image according to the semantic-based illumination color feature and the statistical rule-based illumination color feature.
1. A method for performing illumination color prediction on an image in a neural network model, comprising:
inputting an image to the neural network model;
extracting a semantic-based illumination color feature of the image and a statistical rule-based illumination color feature of the image; and
predicting an illumination color of the image according to the semantic-based illumination color feature and the statistical rule-based illumination color feature,
wherein, when the statistical rule-based illumination color feature is extracted, a spatial correlation between adjacent pixel points in the image is broken.
2. The method according to claim 1 , wherein extracting the statistical rule-based illumination color feature of the image comprises of:
dividing a plurality of pixel blocks randomly from a feature map of the image; and
obtaining the statistical rule-based illumination color feature of the image by integrating the divided pixel blocks.
3. The method according to claim 2 , wherein integrating the divided pixel blocks comprises of:
obtaining an integrated pixel block by adding or multiplying or averaging pixel values of pixel points of the divided pixel blocks at a same position.
4. The method according to claim 2 , wherein a ratio of a number of pixel points divided into the pixel blocks to a number of all pixel points in the feature map of the image is not less than a predetermined ratio.
5. The method according to claim 2 , wherein pixel points between a plurality of the divided pixel blocks do not overlap at all, or there are partially or completely overlapped pixel blocks.
6. An image processing method comprising:
predicting an illumination color of an image according to a semantic-based illumination color feature and a statistical rule-based illumination color feature;
removing the predicted illumination color from the image; and
performing a target recognition on the image from which the illumination color is removed,
wherein, when the statistical rule-based illumination color feature is extracted, a spatial correlation between adjacent pixel points in the image is broken.
7. An image processing method comprising:
predicting an illumination color of an image collected in real time in a viewfinder according to a semantic-based illumination color feature and a statistical rule-based illumination color feature; and
automatically adjusting an exposure amount of capturing according to the predicted illumination color,
wherein, when the statistical rule-based illumination color feature is extracted, a spatial correlation between adjacent pixel points in the image is broken.
8. An image processing method comprising:
predicting an illumination color of an image according to a semantic-based illumination color feature and a statistical rule-based illumination color feature; and
changing the predicted illumination color to other different illumination color to obtain an image after the illumination color is changed,
wherein, when the statistical rule-based illumination color feature is extracted, a spatial correlation between adjacent pixel points in the image is broken.
9. An apparatus for performing illumination color prediction on an image in a neural network model, comprising:
at least one processor; and
a memory that stores a program which, when executed by the at least one processor, causes the processor to function as:
an input unit configured to input an image to the neural network model;
an extraction unit configured to extract a semantic-based illumination color feature of the image and a statistical rule-based illumination color feature of the image; and
a prediction unit configured to predict an illumination color of the image based on the semantic-based illumination color feature and the statistical rule-based illumination color feature,
wherein, when the statistical rule-based illumination color feature is extracted, a spatial correlation between adjacent pixel points in the image is broken.
10. The apparatus according to claim 9 , wherein the extraction unit comprises of:
a divider configure to divide a plurality of pixel blocks randomly from a feature map of the image; and
an obtaining unit configured to obtain the statistical rule-based illumination color feature of the image by integrating the divided pixel blocks.
11. The apparatus according to claim 10 , the obtaining unit further configured to obtain an integrated pixel block by adding or multiplying or averaging pixel values of pixel points of the divided pixel blocks at a same position.
12. The apparatus according to claim 10 , wherein a ratio of a number of pixel points divided into the pixel blocks to a number of all pixel points in the feature map of the image is not less than a predetermined ratio.
13. An image processing apparatus, comprising:
at least one processor; and
a memory that stores a program which, when executed by the at least one processor, causes the processor to function as:
a prediction unit configured to predict an illumination color of an image based on an extracted semantic-based illumination color feature and an extracted statistical rule-based illumination color feature;
an illumination color removing unit configured to remove, from the image, the illumination color predicted by the prediction unit; and
a recognition unit configured to perform a target recognition on the image from which the illumination color is removed,
wherein, when the statistical rule-based illumination color feature is extracted, a spatial correlation between adjacent pixel points in the image is broken.
14. An image processing apparatus, comprising:
at least one processor; and
a memory that stores a program which, when executed by the at least one processor, causes the processor to function as:
a prediction unit configured to predict an illumination color of an image based on an extracted semantic-based illumination color feature and an extracted statistical rule-based illumination color feature; and
an exposure amount adjusting unit configured to automatically adjust an exposure amount of capturing, according to an illumination color of an image collected in real time in a viewfinder predicted by the prediction unit;
wherein, when the statistical rule-based illumination color feature is extracted, a spatial correlation between adjacent pixel points in the image is broken.
15. An image processing apparatus, comprising:
at least one processor; and
a memory that stores a program which, when executed by the at least one processor, causes the processor to function as:
a prediction unit configured to predict an illumination color of an image based on an extracted semantic-based illumination color feature and an extracted statistical rule-based illumination color feature; and
an illumination color changing unit configured to change the illumination color predicted by the prediction unit to other different illumination color to obtain an image after the illumination color is changed,
wherein, when the statistical rule-based illumination color feature is extracted, a spatial correlation between adjacent pixel points in the image is broken.
16. A non-transitory computer-readable storage medium, on which a computer program is stored, and the computer program, when being executed by a processor, causes the processor to execute a method, the method comprising:
inputting an image to a neural network model;
extracting a semantic-based illumination color feature of the image and a statistical rule-based illumination color feature of the image; and
predicting an illumination color of the image according to the semantic-based illumination color feature and the statistical rule-based illumination color feature,
wherein, when the statistical rule-based illumination color feature is extracted, a spatial correlation between adjacent pixel points in the image is broken.
17. The non-transitory computer-readable storage medium according to claim 16 , wherein extracting the statistical rule-based illumination color feature of the image comprises of:
dividing a plurality of pixel blocks randomly from a feature map of the image; and
obtaining the statistical rule-based illumination color feature of the image by integrating the divided pixel blocks.
18. The non-transitory computer-readable storage medium according to claim 17 , wherein integrating the divided pixel blocks comprises of:
obtaining an integrated pixel block by adding or multiplying or averaging pixel values of pixel points of the divided pixel blocks at a same position.
19. The non-transitory computer-readable storage medium according to claim 17 , wherein a ratio of a number of pixel points divided into the pixel blocks to a number of all pixel points in the feature map of the image is not less than a predetermined ratio.
20. The non-transitory computer-readable storage medium according to claim 17 , wherein pixel points between a plurality of the divided pixel blocks do not overlap at all, or there are partially or completely overlapped pixel blocks.