Image scaling method and apparatus
View Patent ↗The present invention relates to image processing technologies, and discloses an image scaling method and apparatus to reduce distortion of a scaled image. The image scaling method includes: determining a distribution direction of main objects in a source image; and scaling the source image to a target image through a nonlinear scaling method according to the distribution direction of the main objects in the source image, where the nonlinear scaling method employs a scaling direction vertical to the distribution direction of the main objects in the source image. The embodiments of the present invention are primarily applicable to the image scaling field.
1. An image scaling method, comprising:
determining a distribution direction of main objects comprising persons, faces of persons or areas of rich textures in a source image through detection of the source image;
scaling the source image to a target image through a nonlinear scaling method according to the distribution direction of the main objects in the source image, wherein the nonlinear scaling method employs a scaling direction perpendicular to the distribution direction of the main objects in the source image,
wherein the distribution direction is one of horizontal and vertical;
wherein determining the distribution direction and scaling the source image are performed in a computer device; and
wherein the determining the distribution direction of the main objects comprises:
detecting the source image through face detection method, edge detection method, or entropy coding method;
if the face detection method is performed, detecting the distribution of the faces of persons in the source image, obtaining the distribution direction of the main objects through the distribution direction of the persons and the symmetry information;
if the edge detection method is performed, calculating the edge distribution of the source image, identifying the areas of rich textures in the source image, determining the distribution direction of the main objects; and
if the entropy coding method is performed, detecting the direction with even distribution of the main objects in the source image, and determining the distribution of the main objects in the source image.
2. The image scaling method according to claim 1 , wherein,
if the distribution direction of the main objects in the source image is determined as horizontal, the step of scaling the source image to the target image through the nonlinear scaling method according to the distribution direction of the main objects in the source image comprises:
dividing the source image into at least two segments vertically;
setting a different scaling factor for each segment in the source image; and
transforming pixels in each segment of the source image into the target image at a corresponding scaling factor.
3. The image scaling method according to claim 2 , wherein,
after dividing the source image into at least two segments vertically, the method further comprises dividing the target image into at least two segments corresponding to the source image.
4. The image scaling method according to claim 2 , wherein setting a different scaling factor for each segment in the source image comprises setting a scaling factor for each segment of the main objects and each segment of trivial objects respectively in the source image.
5. The image scaling method according to claim 2 , wherein transforming the pixels in each segment of the source image into the target image at the corresponding scaling factor comprises:
querying a segment of the source image for at least one source pixel corresponding to each target pixel in each corresponding segment of the target image; and
determining coordinate positions and values of target pixels in the target image according to coordinate positions and values of the source pixels.
6. A non-transitory computer-readable storage medium with an executable program stored thereon, wherein the program instructs a computer device to perform the following steps:
determining a distribution direction of main objects comprising persons, faces of persons or areas of rich textures in a source image through detection of the source image; and
scaling the source image to a target image through a nonlinear scaling method according to the distribution direction of the main objects in the source image, wherein the nonlinear scaling method employs a scaling direction perpendicular to the distribution direction of the main objects in the source image;
wherein the distribution direction is one of horizontal and vertical;
wherein the determining the distribution direction of the main objects comprises:
detecting the source image through face detection method, edge detection method, or entropy coding method;
if the face detection method is performed, detecting the distribution of the faces of persons in the image, obtaining the distribution direction of the main objects through the distribution direction of the persons and the symmetry information;
if the edge detection method is performed, calculating the edge distribution of the source image, identifying the areas of rich textures in the source image, determining the distribution direction of the main objects; and
if the entropy coding method is performed, detecting the direction with even distribution of the main objects in the source image, and determining the distribution of the main objects in the source image.
7. The computer-readable storage medium of claim 6 , wherein, if the distribution direction of the main objects in the source image is determined as horizontal, the step of scaling the source image comprises:
dividing the source image into at least two segments vertically;
setting a different scaling factor for each segment in the source image; and
transforming pixels in each segment of the source image into the target image at a corresponding scaling factor.
8. The computer-readable storage medium of claim 7 , wherein, after dividing the source image into at least two segments vertically, the program further instructs the computer device to divide the target image into at least two segments corresponding to the source image.
9. The computer-readable storage medium of claim 7 , wherein the step of setting a different scaling factor for each segment in the source image comprises setting a scaling factor for each segment of the main objects and each segment of trivial objects respectively in the source image.
10. The computer-readable storage medium of claim 7 , wherein the step of transforming the pixels in each segment of the source image into the target image at the corresponding scaling factor comprises:
querying a segment of the source image for at least one source pixel corresponding to each target pixel in each corresponding segment of the target image; and
determining coordinate positions and values of target pixels in the target image according to coordinate positions and values of the source pixels.