Image processing method and apparatus, and shooting terminal
View Patent ↗An image processing method and apparatus, and a shooting terminal, where the method includes acquiring a to-be-processed image that is shot but is not image-processed; extracting shooting characteristic information of the to-be-processed image; performing image processing on the to-be-processed image according to the shooting characteristic information. In this way, after acquiring a to-be-processed image that is shot, a shooting terminal uses shooting characteristic information of the to-be-processed image as a basis of adaptive processing of the image, and performs processing on each to-be-processed image according to shooting characteristic information of each to-be-processed image, so that the quality of an image that is obtained after being processed by the shooting terminal can be adaptively improved.
1. An image processing method comprising:
acquiring an image;
extracting shooting characteristic information from the image;
performing noise reduction and enhancement processing on luminance information of the image according to shooting scene information and shooting metadata; and
performing noise reduction and enhancement processing on chrominance information of the image according to the shooting scene information and the shooting metadata,
wherein performing noise reduction processing on the luminance information of the image comprises:
performing local absolute difference calculation on the luminance information to acquire a sum of absolute difference (SAD) of neighborhood luminance values of the luminance information;
obtaining a first Gaussian noise reduction filter kernel according to the SAD of neighborhood luminance values, the shooting scene information, and the shooting metadata; and
performing Gaussian noise reduction processing on the luminance information according to the first Gaussian noise reduction filter kernel,
wherein performing enhancement processing on the luminance information of the image comprises:
performing edge-preserving-based filtering decomposition on the luminance information to obtain luminance information of a base layer and luminance information of at least one detail layer;
performing contrast enhancement processing on the luminance information of the base layer according to the shooting scene information and the shooting metadata;
determining at least one gain according to the SAD of neighborhood luminance values, the shooting scene information, and the shooting metadata;
performing detail enhancement processing on the luminance information of the detail layer using the gain; and
performing summation on the luminance information that is of the base layer and has undergone contrast enhancement processing and the luminance information that is of the detail layer and has undergone detail enhancement processing, so as to obtain the luminance information that has undergone enhancement processing.
2. The method according to claim 1 , wherein before performing enhancement processing on the luminance information of the image, the method further comprises:
multiplying the luminance information that has undergone Gaussian noise reduction processing by a first coefficient α1 to obtain α1 times the luminance information that has undergone Gaussian noise reduction processing, wherein α1 is a numerical value ranging from 0 to 1;
multiplying the luminance information that has not undergone Gaussian noise reduction processing by (1−α1) to obtain (1−α1) times the luminance information that has not undergone Gaussian noise reduction processing; and
performing summation on the α1 times the luminance information that has undergone Gaussian noise reduction processing and the (1−α1) times the luminance information that has not undergone Gaussian noise reduction processing, so as to obtain first luminance information, and wherein performing enhancement processing on luminance information of the image comprises performing enhancement processing on the first luminance information.
3. The method according to claim 1 , wherein performing noise reduction processing on the chrominance information of the image comprises:
performing local absolute difference calculation according to the chrommance information to acquire a SAD of neighborhood chrominance values of the chrominance information;
acquiring a second Gaussian noise reduction filter kernel according to the SAD of neighborhood chrominance values, the shooting scene information and the shooting metadata; and
performing Gaussian noise reduction processing on the chrominance information according to the second Gaussian noise reduction filter kernel.
4. The method according to claim 3 , wherein performing enhancement processing on the chrominance information of the image according to the shooting scene information and the shooting metadata comprises:
combining the chrominance information and the luminance information to obtain composite image information;
acquiring color saturation of the composite image information; and
adjusting the color saturation according to the shooting scene information and the shooting metadata.
5. The method according to claim 4 , wherein before combining the chrominance information and the luminance information, the method further comprises:
multiplying the luminance information that has undergone enhancement processing by a second coefficient α2 to obtain α2 times the luminance information that has undergone enhancement processing, wherein α2 is a numerical value ranging from 0 to 1;
multiplying the luminance information, that has undergone noise reduction processing by (1−α2) to obtain (1−α2) times the luminance information that has undergone noise reduction processing; and
performing summation on the α2 times the luminance information that has undergone enhancement processing and the (1−α2) times the luminance information that has undergone noise reduction processing, so as to obtain second luminance information, and
wherein combining the chrominance information and the luminance information to obtain composite image information comprises combining the chrominance information and the second luminance information to obtain the composite image information.
6. The method according to claim 3 , further comprising performing salt and pepper noise reduction processing on the chrominance information according to the shooting scene information and the shooting metadata before performing the local absolute difference calculation according to the chrominance information.
7. The method according to claim 1 , further comprising performing salt and pepper noise reduction processing on the luminance information according to the shooting scene information and the shooting metadata before performing local absolute difference calculation on the luminance information.
8. An image processing method comprising:
acquiring an image;
extracting shooting characteristic information from the image;
performing noise reduction and enhancement processing on luminance information of the image according to shooting scene information and shooting metadata; and
performing noise reduction and enhancement processing on chrominance information of the image according to the shooting scene information and the shooting metadata,
wherein performing noise reduction and enhancement processing on the luminance information of the image comprises:
performing local absolute difference calculation on the luminance information to acquire a sum of absolute difference (SAD) of neighborhood luminance values of the luminance information;
obtaining a first Gaussian noise reduction filter kernel according to the SAD of neighborhood luminance values, the shooting scene information and the shooting metadata;
performing convolution on a high-pass filter kernel and the first Gaussian noise reduction filter kernel to acquire a convolution filter kernel;
determining a weight of the convolution filter kernel according to the SAD of neighborhood luminance values, the shooting scene information and the shooting metadata;
multiplying the convolution filter kernel by the weight to acquire a sharpness enhancement filter kernel;
performing summation on the first Gaussian noise reduction filter kernel and the sharpness enhancement filter kernel to obtain a Gaussian noise reduction sharpening filter kernel; and
performing Gaussian noise reduction processing and sharpness enhancement processing on the luminance information according to the Gaussian noise reduction sharpening filter kernel.
9. The method according to claim 8 , further comprising performing salt and pepper noise reduction processing on the luminance information according to the shooting scene information and the shooting metadata before performing local absolute difference calculation on the luminance information.
10. The method according to claim 8 , wherein performing noise reduction processing on the chrominance information of the image comprises:
performing local absolute difference calculation according to the chrominance information to acquire a SAD of neighborhood chrominance values of the chrominance information;
acquiring a second Gaussian noise reduction filter kernel according to the SAD of neighborhood chrominance values, the shooting scene information and the shooting metadata; and
performing Gaussian noise reduction processing on the chrominance information according to the second Gaussian noise reduction filter kernel.
11. The method according to claim 10 , wherein performing enhancement processing on the chrominance information of the image according to the shooting scene information and the shooting metadata comprises:
combining the chrominance information and the luminance information to obtain composite image information;
acquiring color saturation of the composite image information; and
adjusting the color saturation according to the shooting scene information and the shooting metadata.
12. The method according to claim 11 , wherein before combining the chrominance information and the luminance information, the method further comprises:
multiplying the luminance information that has undergone enhancement processing by a second coefficient α2 to obtain α2 times the luminance information that has undergone enhancement processing, wherein α2 is a numerical value ranging from 0 to 1;
multiplying the luminance information that has undergone noise reduction processing by (1−α2) to obtain (1−α2) times the luminance information that has undergone noise reduction processing; and
performing summation on the α2 times the luminance information that has undergone enhancement processing and the (1−α2) times the luminance information that has undergone noise reduction processing, so as to obtain second luminance information, and wherein combining the chrominance information and the luminance information to obtain composite image information comprises combining the chrominance information and the second luminance information to obtain the composite image information.
13. The method according to claim 10 , further comprising: performing salt and pepper noise reduction processing on the chrominance information according to the shooting scene information and the shooting metadata before performing the local absolute difference calculation according to the chrominance information.
14. An image processing apparatus, comprising:
a processor; and
a memory storing instructions, which when executed by the processor, cause the apparatus to:
acquire an image;
extract shooting characteristic information from the image;
perform noise reduction and enhancement processing on luminance information of the image according to shooting scene information and shooting metadata; and
perform noise reduction and enhancement processing on chrominance information of the image according to the shooting scene information and the shooting metadata,
wherein instructions causing the apparatus to perform noise reduction processing on the luminance information of the image comprise instructions causing the apparatus to:
perform local absolute difference calculation on the luminance information to acquire a sum of absolute difference (SAD) of neighborhood luminance values of the luminance information;
obtain a first Gaussian noise reduction filter kernel according to the SAD of neighborhood luminance values, the shooting scene information, and the shooting metadata; and
perform Gaussian noise reduction processing on the luminance information according to the first Gaussian noise reduction filter kernel,
wherein instructions causing the apparatus to perform enhancement processing on the luminance information of the image comprise instructions causing the apparatus to:
perform edge-preserving-based filtering decomposition on the luminance information to obtain luminance information of a base layer and luminance information of at least one detail layer;
perform contrast enhancement processing on the luminance information of the base layer according to the shooting scene information and the shooting metadata;
determine at least one gain according to the SAD of neighborhood luminance values, the shooting scene information, and the shooting metadata;
perform detail enhancement processing on the luminance information of the detail layer using the gain; and
perform summation on the luminance information that is of the base layer and has undergone contrast enhancement processing and the luminance information that is of the detail layer and has undergone detail enhancement processing, so as to obtain the luminance information that has undergone enhancement processing.
15. The apparatus according to claim 14 , wherein before performing enhancement processing on the luminance information of the image, the instructions further cause the apparatus to:
multiply the luminance information that has undergone Gaussian noise reduction processing by a first coefficient α1 to obtain α1 times the luminance information that has undergone Gaussian noise reduction processing, wherein α1 is a numerical value ranging from 0 to 1;
multiply the luminance information that has not undergone Gaussian noise reduction processing by (1−α1) to obtain (1−α1) times the luminance information that has not undergone Gaussian noise reduction processing; and
perform summation on the α1 times the luminance information that has undergone Gaussian noise reduction processing and the (1−α1) times the luminance information that has not undergone Gaussian noise reduction processing, so as to obtain first luminance information, and wherein performing enhancement processing on luminance information of the image comprises performing enhancement processing on the first luminance information.
16. The apparatus according to claim 14 , wherein instructions causing the apparatus to perform noise reduction processing on the chrominance information of the image comprise instructions causing the apparatus to:
perform local absolute difference calculation according to the chrominance information to acquire a SAD of neighborhood chrominance values of the chrominance information;
acquire a second Gaussian noise reduction filter kernel according to the SAD of neighborhood chrominance values, the shooting scene information and the shooting metadata; and
perform Gaussian noise reduction processing on the chrominance information according to the second Gaussian noise reduction filter kernel.
17. The apparatus according to claim 16 , wherein instructions causing the apparatus to perform enhancement processing on the chrominance information of the image according to the shooting scene information and the shooting metadata comprise instructions causing the apparatus to:
combine the chrominance information and the luminance information to obtain composite image information;
acquire color saturation of the composite image information; and
adjust the color saturation according to the shooting scene information and the shooting metadata.
18. The apparatus according to claim 17 , wherein before combining the chrominance information and the luminance information, the instructions further cause the apparatus to:
multiply the luminance information that has undergone enhancement processing by a second coefficient α2 to obtain α2 times the luminance information that has undergone enhancement processing, wherein α2 is a numerical value ranging from 0 to 1;
multiply the luminance information that has undergone noise reduction processing by (1−α2) to obtain (1−α2) times the luminance information that has undergone noise reduction processing; and
perform summation on the α2 times the luminance information that has undergone enhancement processing and the (1−α2) times the luminance information that has undergone noise reduction processing, so as to obtain second luminance information, and wherein combining the chrominance information and the luminance information to obtain composite image information comprises combining the chrominance information and the second luminance information to obtain the composite image information.
19. The apparatus according to claim 16 , wherein the instructions further cause the apparatus to perform salt and pepper noise reduction processing on the chrominance information according to the shooting scene information and the shooting metadata before performing the local absolute difference calculation according to the chrominance information.
20. The apparatus according to claim 14 , wherein the instructions further cause the apparatus to perform salt and pepper noise reduction processing on the luminance information according to the shooting scene information and the shooting metadata before performing local absolute difference calculation on the luminance information.