
The "blind deconvolution" method restores the original image and the blurring function in the case that a blurred image is presented and the original image and the blurring function are unknown. This method applies the Fourier transform and the inverse Fourier transform repeatedly to the blurred image, and revises the image during the repetition using constraints like "The area where the image exists should be bounded," "The brightness value of each pixel should be positive," and so on. Our method estimates the original image from a blurred RGB color image using additional constraints on color images. Examples: blurred images (left), and restored images (right).
Examples of blind deconvolution.
