Separating a directional lighting variability in statistical face modelling based on texture space decomposition

1. A method of determining a characteristic of a face within a scene captured in a digital image, comprising:

building a first model from a plurality of images including different individual faces, each acquired independently of directional lighting variations, said model comprising a mean, shape vector (s), a mean texture vector (t), and a first texture vector (b t ) for each face;

building respective second models from respective sets of images, each set acquired for a given individual with varying directional lighting, said respective second models comprising said mean shape vector (s) and respective second texture vectors (g) for said images;

projecting said second texture vectors (g) onto the subspace spanned by the first model to provide a filtered texture vector (g filt )for each image of said sets of images;

subtracting said filtered texture vector (g filt ) from said second texture vectors (g) to provide a residual texture vector (g res ) for each image of said sets of images;

building a second model orthogonal to said first model for said sets of images based on a mean of said residual texture vectors (g res ), said second model including a texture vector (bg) for each image;

fusing said first and second texture (b t , b g ) vectors for each image to provide a fused texture model (t fused );

acquiring a digital image including a face within a scene;

determining an initial location of the face in the scene;

applying said fused model to said initial location to obtain a active-appearance based fit of said fused model to said face;

based on the obtained fit of the fused model to said face in the scene, determining at least one characteristic of the face; and

electronically storing, transmitting, applying a face recognition program to, editing, or displaying the face including the determined characteristic, or combinations thereof.

2. The method of claim 1 , wherein the model components comprise eigenvectors, and the texture values comprises eigenvalues of the eigenvectors.

3. The method of claim 1 , Wherein the at least one determined characteristic comprises a feature that is independent of directional lighting.

4. The method of claim 1 , further comprising generating a reconstructed image from said acquired digital image without periodic noise component, including obtaining a second fit of the face to a second linear texture model that is based on a training data set and that comprises a class of objects including a set of model components which are without a periodic noise component.

5. The method of claim 4 , further comprising:

extracting the periodic noise component including determining a difference between the face and the reconstructed image, and

determining a frequency of the noise component; and

removing the periodic noise component of the determined frequency.

6. The method of claim 1 , further comprising determining an exposure value for the face of said acquired digital image, including obtaining a fit to the face to a second linear texture model that is based on a training data set and that comprises a class of objects including a set of model components that exhibit a dependency on exposure value variations.

7. The method of claim 6 , further comprising including reducing an effect of a background region or density contrast caused by shadow or both.

8. The method of claim 1 , further comprising controlling a flash for acquiring said digital image to accurately reflect a lighting condition, including obtaining a flash control condition by referring to a reference table and controlling a flash light emission according to the flash control condition.

9. The method of claim 8 , further Comprising reducing an effect of contrasting density caused by shadow or black compression or white compression or combinations thereof.

10. The method of claim 1 , further comprising adjusting or determining a sharpness value, or both, of said acquired digital image including:

applying a second linear texture model that is constructed based on a training data set and comprises a class of objects including a subset of model components that exhibit a dependency on sharpness variations,

obtaining a fit of said second model to said face in the scene including adjusting one or more individual values of one or more model components of said second linear texture model; and

based on the obtained fit of the second model to said face in the scene, adjusting a sharpness of the face including changing one or more values of one or more model components of the second linear texture model to generate a further adjusted object model.

11. The method of claim 1 , further comprising removing a blemish from a face of said acquired digital image, including:

applying a second linear texture model that is constructed based on a training data set and comprises a class of objects including a subset of model components that do not include such a blemish;

obtaining a fit of said second model to said face in the scene including adjusting one or more individual values of one or more model components of said second linear texture model; and

based on the obtained fit of the second model to said face in the scene, removing the blemish from the face including changing one or more values of one or more model components of the second linear texture model to generate a further adjusted object model.

12. The method of claim 11 , wherein the blemish comprises an acne blemish or other skin blemish; or a photographic artefact.

13. The method of claim 1 , further comprising adjusting or determining a graininess value, or both, of said acquired digital image including:

applying a second linear texture model that is constructed based on a training data set and comprises a class of objects including a subset of model components that exhibit a dependency on graininess variations,

obtaining a fit of said second model to said face in the scene including adjusting one or more individual values of one or more model components of said second linear texture model; and

based on the obtained fit of the second model to said face in the scene, adjusting a graininess of the face including changing one or more values of one or more model components of the second linear texture model to generate a further adjusted object model.

14. The method of claim 1 , further comprising converting, adjusting or determining a resolution value, or combinations thereof, of said acquired digital image including:

applying a second linear texture model that is constructed based on a training data set and comprises a class of objects including a subset of model components that exhibit approximately a same resolution as said face,

obtaining a fit of said second model to said face in the scene including adjusting one or more individual values of one or more model components of said second linear texture model; and

based on the obtained fit of the second model to said face in the scene, converting a resolution of the face including changing one or more values of one or more model components of the second linear texture model to generate a further adjusted object model.

15. A digital image acquisition device including an optoelectonic system for acquiring a digital image, and a digital memory having stored therein processor-readable code for programming the processor to perform a method of determining a characteristic of a face within a scene captured in a digital image, according to any of claims 1 - 14 .

16. A non-transitory digital memory having stored therein processor-readable code for programming a processor to perform a method of determining a characteristic of a face within a scene captured in a digital image, according to any of claims 1 - 14 .