Bayer-consistent raw scaling

摘要

A system and method for scaling an image includes receiving raw image data comprising input pixel values which correspond to pixels of an image sensor; and filtering pixels according to a Bayer-consistent ruleset by a first scaling factor in a first direction and a second scaling factor in a second direction perpendicular to the first direction, wherein the first scaling factor is different from the second scaling factor. The system and method may also include outputting scaled image data as output pixel values, which correspond to subgroups of the input pixel values. The Bayer-consistent ruleset includes a set of filter weights and a series of scaling rules. The Bayer-consistent ruleset results in a scaled image having a high degree of Bayer-consistency.

主权项

1. A method of scaling an image comprising:
receiving raw image data comprising a plurality of input pixel values, respective ones of the input pixel values corresponding to a respective pixel of an image sensor; and filtering the raw image data according to a Bayer-consistent ruleset by a first scaling factor in a first direction, and a second scaling factor in a second direction perpendicular to the first direction, wherein the first scaling factor is different from the second scaling factor, and the Bayer-consistent ruleset results in a Bayer-consistency coefficient C of greater than or equal to 0.65, where C is defined according to the following expression (1):
C=0.5γ+0.5β  (1), where γ and β are defined according to the following expressions (2) and (3), respectively:γ=14⁢∑c⁢⁢Σc⁢⁢sub⁢-⁢tile⁡(filter⁢⁢coefficients⁢⁢for⁢⁢c⁢⁢colored⁢⁢pixels)2Σall⁢⁢sub⁢-⁢tiles⁡(filter⁢⁢coefficients⁢⁢for⁢⁢c⁢⁢colored⁢⁢pixels)2,(2)⁢β=max⁡(1-14⁢∑c⁢⁢(dchN⁢⁢Δh)2+(dcvM⁢⁢Δv)2,0),(3) where dch and dcv are the horizontal and vertical distances, respectively, between a center of gravity of the filter coefficients for the color c and a geometric pixel center of a corresponding c color output pixel, N is a horizontal scaling factor, M is a vertical scaling factor, Δh is a horizontal length of the input pixel, and Δv is a vertical length of an input pixel.