Separating a directional lighting variability in statistical face modelling based on texture space decomposition
First Claim
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 (bt) 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 (gfilt)for each image of said sets of images;
subtracting said filtered texture vector (gfilt) from said second texture vectors (g) to provide a residual texture vector (gres) 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 (gres), said second model including a texture vector (bg) for each image;
fusing said first and second texture (bt, bg) vectors for each image to provide a fused texture model (tfused);
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.
4 Assignments
0 Petitions
Accused Products
Abstract
A technique for determining a characteristic of a face or certain other object within a scene captured in a digital image including acquiring an image and applying a linear texture model that is constructed based on a training data set and that includes a class of objects including a first subset of model components that exhibit a dependency on directional lighting variations and a second subset of model components which are independent of directional lighting variations. A fit of the model to the face or certain other object is obtained including adjusting one or more individual values of one or more of the model components of the linear texture model. Based on the obtained fit of the model to the face or certain other object in the scene, a characteristic of the face or certain other object is determined.
-
Citations
16 Claims
-
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 (bt) 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 (gfilt)for each image of said sets of images; subtracting said filtered texture vector (gfilt) from said second texture vectors (g) to provide a residual texture vector (gres) 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 (gres), said second model including a texture vector (bg) for each image; fusing said first and second texture (bt, bg) vectors for each image to provide a fused texture model (tfused); 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. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16)
-
Specification