Block-matching algorithm for color interpolation
AssigneeIntel Corporation
InventorTinku Acharya
Life of this patent
Estimating the term…
Abstract
What is disclosed is method that interpolates missing color components for an original pixel by utilizing color gravities for surrounding blocks of green pixels. A new pixel is output which has complete color information which is a function of the original pixel and interpolated color components. The interpolated pixel is a true representation of both luminance and chrominance information.
Claims41
A method for color interpolation of a raw image composed of pixels, comprising:
- interpolating a missing color component for each raw image pixel by selecting from neighboring pixels based on color gravities for blocks of Green pixels surrounding the neighboring pixels;
- and
- forming a new full color pixel having full color information, the full color information being a function of the raw image pixel and of the interpolated missing color component, wherein the interpolating comprises:
- determining if the raw image pixel is Green;
- if the raw image pixel is Green, applying a first interpolation methodology for interpolating missing Red and Blue components;
- and
- if the raw image pixel is not Green, applying a second interpolation methodology for interpolating a missing Green color component and one of missing Blue and Red color component.
An imaging apparatus comprising:
- an image processing circuit to receive raw image data, said image processing unit configured to interpolate color components missing in said raw image data by selecting neighboring pixels based on color gravities of blocks of Green pixels surrounding said neighboring pixels.
A method of color interpolation for a raw image composed of pixels, comprising:
- interpolating a missing color component for a raw image pixel by selecting from neighboring pixels based on color gravities for blocks of pixels of a first color surrounding the neighboring pixels, a selected pixel being set as an interpolated missing color component in the case of missing color components other than of the first color;
- and
- forming a new pixel as having full color information, the full color information being a function of the raw pixel and of the interpolated missing color components, where the interpolation includes:
- determining if the raw image pixel is of the first color;
- if the raw image pixel is of the first color, applying a first interpolation methodology to interpolate missing components of second and third colors;
- and
- if the raw image pixel is not of the first color, applying a second interpolation methodology to interpolate a missing component of the first color and a missing component of one of the second color and the third color.
A machine readable medium having instructions stored thereon that when executed by a processor cause a system to:
- interpolate a missing color component for a raw image pixel by selecting from neighboring pixels based on color gravities for blocks of pixels of a first color surrounding the neighboring pixels, a selected pixel being set as an interpolated missing color component in the case of missing color components other than of the first color;
- and
- form a new pixel as having full color information, the full color information being a function of the raw pixel and of the interpolated missing color component, where the interpolation includes determining if the raw image pixel is of the first color, if the raw image pixel is of the first color, applying a first interpolation methodology to interpolate missing components of second and third colors, and if the raw image pixel is not of the first color, applying a second interpolation methodology to interpolate a missing component of the first color and a missing component of one of the second color and the third color.
A machine readable medium having instructions stored thereon that when executed by a processor cause a system to perform operations comprising:
- interpolating a missing color component for each raw image pixel by selecting from neighboring pixels based on color gravities for blocks of Green pixels surrounding the neighboring pixels;
- and
- forming a new full color pixel having full color information, the full color information being a function of the raw image pixel and of the interpolated missing color component, wherein interpolating comprises:
- determining if the raw image pixel is Green;
- if the raw image pixel is Green, applying a first interpolation methodology for interpolating missing Red and Blue components;
- and
- if the raw image pixel is not Green, applying a second interpolation methodology for interpolating a missing Green component and one of missing Blue and Red component.
An apparatus comprising:
- a camera having image processing circuitry to compress digital images captured by an image sensor in said camera, and an image memory unit to store the compressed images, the digital images being derived from raw images captured by the sensor and having full color information, the image processing circuit to perform color interpolation on a raw pixel by setting a missing component of said raw pixel to be a value of a selected neighbor raw pixel that has the same color as that of the missing component, based on a comparison involving color gravities of a first pixel block surrounding said selected neighbor raw pixel and of a second pixel block surrounding another same color neighbor raw pixel.
A method for color interpolation of a digital image having raw pixels, comprising:
- setting a missing non-green component of a raw pixel to be a value of a selected neighbor raw pixel that has the same color as that of the missing component, based on a comparison involving color gravities of a first pixel block surrounding said selected neighbor raw pixel and of a second pixel block surrounding another same color neighbor raw pixel.
Dependencies are inferred from the claim text.