METHODS AND SYSTEMS OF COMPARING FACE MODELS FOR RECOGNITION

• US 20090244082A1
• Filed: 04/01/2009
• Published: 10/01/2009
• Est. Priority Date: 04/01/2008
• Status: Active Grant

• Current Assignee: The Navy Chief Of Naval Research Office Of Counsel
• Original Assignee: The Navy Chief Of Naval Research Office Of Counsel
• Inventors: Livingston, Mark A., Solomon, Justin
• Granted Patent: US 8,477,147 B2
• Time in Patent Office: Days
• US Class Current: 345/581
• CPC Class Codes: G06V 10/426 Graphical representations
• CPC Class Codes: G06V 10/443 by matching or filtering
• CPC Class Codes: G06V 40/169 Holistic features and representations, i.e. based on the facial image taken as a whole
• CPC Class Codes: G06V 40/171 Local features and components; Facial parts eye characteristics G06V40/18; Occluding parts, e.g. glasses; Geometrical relationships

Abstract

Methods and systems of representation and manipulation of surfaces with perceptual geometric features, using a computer graphics rendering system, include executing algorithmic instructions to compute a plurality of vertices, edges and surfaces in a mesh for the purpose of defining representations of surfaces on grids. Normals and distances are determined for triangular surfaces to be considered. Additionally, height fields of a function are defined. A set of feature curves and a set of feature points are derived, based on the defined function. Infinitesimal movements along the representations of the surfaces are determined, along with derivations of properties of representations of continuous surfaces. Additional determinations of perceptual geometric features include determinations such as zero crossings, parabolic curves, flecnodes, ruffles, gutterpoints, conical points and biflecnodes in a given mesh. After these determinations are made, visual representation are rendered which captures perceptually important features for smoothly varying shapes.

Claims

1. A method of representation and manipulation of surfaces with perceptual geometric features using a computer processor, the method comprising:

receiving a request to perform operations of representation and manipulation of surfaces;

initiating the computer processor to execute visualization of surfaces computer readable and computer executable program code;

describing a first plurality of geometric models with a first plurality of geometric surface features significant to a human visual system;

detecting a second plurality of geometric surface features on a second plurality of existing geometric models which are comparable to the first plurality of geometric surface features significant to the human visual system described in the first plurality of geometric models, wherein the first plurality of geometric surface features significant to the human visual system are psychophysics geometric surface features;

comparing the first plurality of geometric surface features significant to the human visual system with the second plurality of geometric surface features, wherein comparing includes analyzing perceptual geometric features by considering the plurality of psychophysics feature models favoring the human visual system;

filling in features, between the first plurality of geometric surface features and the second plurality of geometric features, using a model matching algorithm to create one of a partial model and an entire model of new features;

rendering, by the computer processor using the model matching algorithm, comparisons of geometric surface features of the plurality of existing psychophysics feature models and perceptual geometric features and one of the partial model and the entire model of new features; and

applying results of analyzing, determining and rendering, to tasks in which one of shapes and spatial relations must be analyzed, recognized and understood, in regard to one of face visualization models and terrain visualization models.

2. The method of claim 1, wherein filing in features includes:

determining shape transformation, erosion modeling and morphing, shape matching, by using the model matching algorithm to perform operations including;

computing a surface normal;

considering a plurality of vertices, edges and surfaces in a mesh;

defining a representation on one of a regular grid and an irregular grid;

determining a normal of each surface which is a triangle of a plurality of triangles;

defining a distance of a vertex for each triangle of the plurality of triangles;

defining a height field of a function normal;

deriving a set of feature curves and a set of feature points;

determining an infinitesimal movement along representations of surfaces;

deriving properties of representations of continuous surfaces;

determining zero crossings in curvatures, when representations of surfaces change between one of convex-hyperbolic and concave-hyperbolic;

determining parabolic curves by traversing the mesh until an edge is found, wherein a curvature is negative at a first endpoint and positive at a second endpoint;

determining a plurality of flecnodes;

determining ruffles;

determining gutterpoints;

locating biflecnodes;

determining a conical point; and

rendering visualization of surfaces with perceptual geometric features using a computer processor communicatively connected to a graphics output device.

3. The method of claim 1, wherein filing in features further includes:

determining shape transformation, erosion modeling and morphing, shape matching, by using the model matching algorithm, wherein the model matching algorithm is a face matching algorithm.

4. The method of claim 3, wherein the face matching algorithm performs operations including:

inputting a height field mesh of a face;

identifying convex, concave and hyperbolic regions of the height field mesh of the face;

computing properties for convex, concave and hyperbolic regions;

constructing an association graph for two face meshes to be compared;

creating a node in the association graph for two face meshes for each member of a Cartesian product of nodes;

creating an edge, in the association graph for two face meshes, connecting antecedent nodes;

finding a size of fully connected antecedent nodes;

decomposing surfaces into patches; and

rendering visualization of surfaces with perceptual geometric features using a computer processor communicatively connected to a graphics output device.

5. A system of providing representation and manipulation of surfaces with perceptual geometric features, the system comprising:

a computer processor communicatively coupled to an input output device; and

a memory, residing in the computer processor, having a dynamic repository and a program unit containing a computer readable and computer executable program, wherein the computer readable and computer executable program residing in the program unit, when executed by the computer processor causes the computer processor to perform operations of;

receiving a request to perform operations of visualization of surfaces;

initiating the computer processor to execute representation and manipulation of surfaces operations computer readable and computer executable program code;

describing a first plurality of geometric models with a first plurality of geometric surface features significant to a human visual system;

detecting a second plurality of geometric surface features on a second plurality of existing geometric models which are comparable to the first plurality of geometric surface features significant to the human visual system described in the first plurality of geometric models, wherein the first plurality of geometric surface features significant to the human visual system are psychophysics geometric surface features;

comparing the first plurality of geometric surface features significant to the human visual system with the second plurality of geometric surface features, wherein comparing includes analyzing perceptual geometric features by considering the plurality of psychophysics feature models favoring the human visual system;

filling in features, between the first plurality of geometric surface features and the second plurality of geometric features, using a model matching algorithm to create one of a partial model and an entire model of new features;

rendering, by the computer processor using the model matching algorithm, comparisons of geometric surface features of the plurality of existing psychophysics feature models and perceptual geometric features and one of the partial model and the entire model of new features; and

applying results of analyzing, determining and rendering, to tasks in which one of shapes and spatial relations must be analyzed, recognized and understood, in regard to one of face visualization models and terrain visualization models.

6. The system of claim 5, wherein filing in features includes the computer processor executing the model matching algorithm to perform operations including:

determining shape transformation, erosion modeling and morphing, shape matching.

7. The system of claim 6, wherein determining shape transformation erosion modeling and morphing, shape matching includes having the computer processor executing the model matching algorithm to perform operations including:

computing a surface normal;

considering a plurality of vertices, edges and surfaces in a mesh;

defining a representation on one of a regular grid and an irregular grid;

determining a normal of each surface which is a triangle of a plurality of triangles;

defining a distance of a vertex for each triangle of the plurality of triangles;

defining a height field of a function normal;

deriving a set of feature curves and a set of feature points;

determining an infinitesimal movement along representations of surfaces;

deriving properties of representations of continuous surfaces;

determining zero crossings in curvatures, when representations of surfaces change between one of convex-hyperbolic and concave-hyperbolic;

determining parabolic curves by traversing the mesh until an edge is found, wherein a curvature is negative at a first endpoint and positive at a second endpoint;

determining a plurality of flecnodes;

determining ruffles;

determining gutterpoints;

locating biflecnodes;

determining a conical point; and

rendering visualization of surfaces with perceptual geometric features using a computer processor communicatively connected to a graphics output device.

8. The system of claim 7, wherein the model matching algorithm is a face matching algorithm.

9. The system of claim 8, wherein the computer processor executing the face matching algorithm performs operations including:

inputting a height field mesh of a face;

identifying convex, concave and hyperbolic regions of the height field mesh of the face;

computing properties for convex, concave and hyperbolic regions;

constructing an association graph for two face meshes to be compared;

creating a node in the association graph for two face meshes for each member of a Cartesian product of nodes;

creating an edge, in the association graph for two face meshes, connecting antecedent nodes;

finding a size of fully connected antecedent nodes;

decomposing surfaces into patches; and

rendering visualization of surfaces with perceptual geometric features using a computer processor communicatively connected to a graphics output device.

10. A computer readable medium having a plurality of computer readable and computer executable instructions of a method of representation and manipulation of surfaces with perceptual geometric features when executed on a computer processor, causes the computer processor to execute computer readable and computer executable instructions to perform operations of a method comprising:

instructions initiating the computer processor to execute representation and manipulation of surfaces operations computer readable and computer executable program code;

instructions describing a first plurality of geometric models with a first plurality of geometric surface features significant to a human visual system;

instructions detecting a second plurality of geometric surface features on a second plurality of existing geometric models which are comparable to the first plurality of geometric surface features significant to the human visual system described in the first plurality of geometric models, wherein the first plurality of geometric surface features significant to the human visual system are psychophysics geometric surface features;

instructions comparing the first plurality of geometric surface features significant to the human visual system with the second plurality of geometric surface features, wherein comparing includes analyzing perceptual geometric features by considering the plurality of psychophysics feature models favoring the human visual system;

instructions filling in features, between the first plurality of geometric surface features and the second plurality of geometric features, using a model matching algorithm to create one of a partial model and an entire model of new features;

instructions rendering, by the computer processor using the model matching algorithm, comparisons of geometric surface features of the plurality of existing psychophysics feature models and perceptual geometric features and one of the partial model and the entire model of new features; and

instructions applying results of analyzing, determining and rendering, to tasks in which one of shapes and spatial relations must be analyzed, recognized and understood, in regard to one of face visualization models and terrain visualization models.

11. The computer readable medium of claim 10, wherein instructions filling in features using the model matching algorithm includes:

instructions determining shape transformation, erosion modeling and morphing, shape matching.

12. The computer readable medium of claim 11, wherein instructions of the model matching algorithm which include instructions determining shape transformation, erosion modeling and morphing, shape matching further have instructions including:

instructions computing a surface normal;

instructions considering a plurality of vertices, edges and surfaces in a mesh;

instructions defining a representation on one of a regular grid and an irregular grid;

instructions determining a normal of each surface which is a triangle of a plurality of triangles;

instructions defining a distance of a vertex for each triangle of the plurality of triangles;

instructions defining a height field of a function normal, characterized as

$N={\frac{1}{\sqrt{1+f_u^2+f_v^2}}<br><br>\left[-f_u-f_v<br><br>1\right]}^T;<br><br>$

instructions deriving a set of feature curves and a set of feature points;

instructions determining an infinitesimal movement along representations of surfaces;

instructions deriving properties of representations of continuous surfaces;

instructions determining zero crossings in curvatures, when representations of surfaces change between one of convex-hyperbolic and concave-hyperbolic;

instructions determining parabolic curves by traversing the mesh until an edge is found, wherein a curvature is negative at a first endpoint and positive at a second endpoint;

instructions determining a plurality of flecnodes;

instructions determining ruffles;

instructions determining gutterpoints;

instructions locating biflecnodes;

instructions determining a conical point; and

instructions rendering visualization of surfaces with perceptual geometric features using a computer processor communicatively connected to a graphics output device.

13. The computer readable medium of claim 12, wherein the model matching algorithm is a face matching algorithm further having a plurality of face matching algorithm instructions.

14. The computer readable medium of claim 13, wherein the plurality of face matching algorithm instructions include:

instructions of inputting a height field mesh of a face;

instructions identifying convex, concave and hyperbolic regions of the height field mesh of the face;

instructions for computing properties for convex, concave and hyperbolic regions;

instructions of constructing an association graph for two face meshes to be compared;

instructions for creating a node in the association graph for two face meshes for each member of a Cartesian product of nodes;

instructions for creating an edge, in the association graph for two face meshes, connecting antecedent nodes;

instructions for finding a size of fully connected antecedent nodes;

instructions decomposing surfaces into patches; and

instructions rendering visualization of surfaces with perceptual geometric features using a computer processor communicatively connected to a graphics output device.