Image correction method, device and terminal

Abstract

An image correction method, device, and terminal. The image correction method includes: sampling an output pixel matrix of a corrected image to obtain output coordinates of a plurality of sampled pixel points, where the output coordinates refer to coordinates in the output pixel matrix; The output coordinates of the sampling pixels, calculate the original coordinates of multiple sampling pixels, the original coordinates are the coordinates in the original pixel matrix, the original pixel matrix is ​​included in the original image; calculate the difference between the original coordinates and the output coordinates of multiple sampling pixels value to obtain the first difference; perform interpolation calculation on the first difference to obtain the difference between the original coordinates of the non-sampled pixels and the output coordinates to obtain the second difference, and the non-sampled pixels are the output pixel matrix pixels other than the sampling pixels in the middle; obtain pixel values ​​of all pixels in the output pixel matrix based on the original pixel matrix, the first difference value and the second difference value. The technical scheme of the invention improves the efficiency of image correction.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to an image correction method, device and terminal. Background technique [0002] In the field of computer vision, in the research fields of image stitching, video stabilization, etc., in order to achieve the purpose of image content alignment and reduce video jitter, it is usually necessary to perform some kind of matrix transformation on the image. Align one image to another through a transformation matrix, and transformations such as translation, rotation, and perspective will occur between the two images. For example, in the video image stabilization operation, the correction matrix (for example, perspective transformation) of each frame image is calculated, and then the corresponding matrix transformation is performed on the frame image to achieve the purpose of image correction and video image stabilization. [0003] In the prior art, among methods for image ...

AI Technical Summary

Problems solved by technology

[0008] However, for a larger-sized image, if software is used to perform pixel-by-pixel transformation, it will increase the bandwidth of the image processing system and affect the performance of the system; while using hardware to perform matrix transformation operations on the image, for example, as shown in the above formula , there are 4 multiplications and 1 division operation. The division operation is a very complex logic in hardware implementation, which increases the complexity of hardware implementation

Images