| 摘要 |
A method for generating a super-resolution image from at least two input images, represented by digital samples. The super-resolution image having a final resolution higher than an initial resolution of one of at least two input images. The method including obtaining, for each input image of the at least two input images, a distortion model to register the input image on a reference grid at the final resolution. The method including warping each input image into a warped image according to the distortion model. Computing warped coordinates respectively for digital samples of the input image. Wherein, the contribution of each digital sample of the input image to the warped image is weighted. The weight applied to each digital sample depending on the warped coordinates of the digital sample. Combining the warped images in order to generate a combined warped image having the final resolution. |
