Computer based digital image noise reduction method based on overlapping planar approximation

摘要

The method of the present invention reduces noise in digital photographic images by decomposing the digital image into smooth regions and edge regions and by electronically blurring the smooth regions. Smooth regions are areas lacking any sharp detail, such as blue sky. Edge regions are regions containing sharp detail, such as grass. The method operates by choosing a set of neighborhood pixel types from a matrix of pixels representing an image. For each target pixel in the image, all neighborhoods that include the target pixel are considered and a linear least squares regression is computed to fit a plane to the code values of the pixels in each neighborhood, resulting in a goodness of fit and a least squares estimate for the code value at the target pixel. Finally, a noise reduced code value for the target pixel is computed as a normalized, weighted sum of the least squares estimates for the target pixel, summed over all the neighborhoods containing the target pixel. The weights are determined as a function of the goodness of fit and the neighborhood type.