Facial image processing method and apparatus
First Claim
1. A facial image processing method for extracting facial features and outputting the corresponding coordinate data based on facial image data, wherein:
- a specific point is defined in the facial image data for each of the right and left eyes and the mouth;
a searching range is set for each facial feature to be extracted, including the right and left eyes and the mouth, based on the coordinate data of at least one of said specific points and referenced to the specific point specified for that facial feature for which the searching range is being set,where the x-coordinate axis represents the side-to-side direction of the facial image and the y-coordinate axis represents the vertical direction of the facial image;
an area of each of the right and left eyes and the mouth is extracted within the corresponding searching range; and
it is determined whether the extracted mouth area is correct by determining whether a line connecting right and left end points in the x-axis direction of the mouth area is greater than a predetermined angle relative to a line through a particular point in each of the right and left eyes.
2 Assignments
0 Petitions
Accused Products
Abstract
A method and apparatus for appropriately and without error extracting specific feature areas of a facial image, and outputting accurate coordinate data for the extracted facial features for use in such applications as three-dimensional morphing, image blending, or facial image identification. Preferably, the inclusion of a searching range setting unit for each facial feature to be extracted develop a searching range for extracting each facial feature referenced to individual points specified by the user for the mouth and left and right eyes based on the coordinate data of those specified points. The areas of those facial features are then extracted from within the defined searching ranges by area extractors, and the coordinate data of the contour points forming the contours of those areas is extracted by contour point extractors. Based on the coordinate values of the specific point specified for each facial feature area, extraction content evaluators then evaluate the relative positions of the extracted areas to determine whether said areas were correctly extracted.
-
Citations
56 Claims
-
1. A facial image processing method for extracting facial features and outputting the corresponding coordinate data based on facial image data, wherein:
-
a specific point is defined in the facial image data for each of the right and left eyes and the mouth; a searching range is set for each facial feature to be extracted, including the right and left eyes and the mouth, based on the coordinate data of at least one of said specific points and referenced to the specific point specified for that facial feature for which the searching range is being set, where the x-coordinate axis represents the side-to-side direction of the facial image and the y-coordinate axis represents the vertical direction of the facial image; an area of each of the right and left eyes and the mouth is extracted within the corresponding searching range; and it is determined whether the extracted mouth area is correct by determining whether a line connecting right and left end points in the x-axis direction of the mouth area is greater than a predetermined angle relative to a line through a particular point in each of the right and left eyes. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8)
-
-
9. A facial image processing method for extracting facial features and outputting the corresponding coordinate data based on facial image data, wherein:
-
a specific point is defined in the facial image data for each of the right and left eyes and the mouth; a searching range for extracting the areas of said facial features is set for each facial feature to be extracted, including the right and left eyes, the mouth and at least one facial contour, based on the coordinate data of at least one of said specific points and referenced to the specific point specified for that facial feature for which the searching range is being set, where the x-coordinate axis represents the side-to-side direction of the facial image and the y-coordinate axis represents the vertical direction of the facial image; the areas of said facial features are extracted within the set searching ranges; and the coordinate data of contour points forming each facial feature contour is extracted from the facial feature areas; and the relative positions of the extracted feature images are evaluated based on the coordinate values of specific points in the facial feature areas to determine whether the facial features were correctly extracted, wherein the evaluation process for determining correct extraction of the mouth area includes determining whether a line connecting right and left end points in the x-axis direction of the mouth area is greater than a predetermined angle relative to a line through a particular point in each of the right and left eyes. - View Dependent Claims (11, 12)
-
-
10. A facial image processing method for extracting facial features and outputting the corresponding coordinate data based on facial image data, wherein:
-
a specific point is defined in the facial image data for each of plural facial features to be extracted; a searching range for extracting the areas of said facial features is set for each facial feature to be extracted based on the coordinate data of said specific points and referenced to the specific point specified for each facial feature; the areas of said facial features are extracted within the set searching ranges; the coordinate data of the contour points forming each facial feature contour is extracted from the facial feature areas; and the relative positions of the extracted feature images are evaluated based on the coordinate values of specific points in the facial feature areas to determine whether the facial features were correctly extracted; wherein the evaluation process for determining correct extraction of the mouth area extracted in the searching range set for extracting the mouth area; compares the y-axis coordinates of the right and left end points in the x-axis direction of the extracted mouth area with the y-axis coordinate of the midpoint in the x-axis direction at the top edge of the extracted mouth area where the x-coordinate axis represents the side-to-side direction of the facial image and the y-coordinate axis represents the vertical direction of the facial image; and determines the extracted mouth area to represent an open mouth when the y-axis coordinate of the midpoint in the x-axis direction is smaller than the y-axis coordinate of either right or left end point in the x-axis direction of said mouth area, determines the extracted mouth to be the lower lip only, and sets a searching range for extracting the upper lip.
-
-
13. A facial image processing method for extracting facial features and outputting the corresponding coordinate data based on facial image data, wherein
the area of each facial feature is extracted using a coordinate system wherein the x-coordinate axis represents the side-to-side direction of the facial image and the y-coordinate axis represents the vertical direction of the facial image, the contour points forming the contours of the eyes and mouth in the corresponding facial feature areas extracted by the facial feature extracting means are obtained, the coordinates of the right and left end points in the x-axis direction of said facial feature areas are then obtained from said contour points, a searching range with a width in the x-axis direction equal to a known 1/n fraction of the distance between the right and left end points of the respective facial feature area is set in each area near the midpoint between said right and left end points, and the coordinates of the top and bottom end points in the y-axis direction of the eye or mouth facial feature areas are obtained within said searching range.
-
15. A facial image processing method for extracting facial features and outputting the corresponding coordinate data based on facial image data, wherein:
-
after extracting the areas of the facial features, the relative positions of the extracted areas are evaluated based on the coordinate values of a specific point in each facial feature area to determine whether the facial feature areas were correctly extracted; and when the extracted area is the mouth area, it is determined whether the extracted area is correct by determining whether a line connecting the right and left end points in the x-axis direction of the mouth area is sloped greater than a predetermined angle relative to a line through a specific point in each of the right and left eyes, where the x-coordinate axis represents the side-to side direction of the facial image and the y-coordinate axis represents the vertical direction of the facial image.
-
-
16. A facial image processing method for extracting facial features and outputting the corresponding coordinate data based on facial image data, wherein:
-
after extracting the areas of the facial features, the relative positions of the extracted areas are evaluated based on the coordinate values of a specific point in each facial feature area to determine whether the facial feature areas were correctly extracted; and when the extracted area is a facial contour area, the plural contour points forming the facial contour are processed in units of three adjacent contour points, the coordinates of the middle of each three contour points are compared with the average of the coordinates of the two adjacent contour points, and the average of the coordinates of the middle contour point if the position of the substituted for the coordinates of the middle contour point if the position of the middle contour point is determined to be to the inside of the face relative to the adjacent contour points, where the x-coordinate axis represents the side-to-side direction of the facial image and the y-coordinate axis represents the vertical direction of the facial image.
-
-
17. A facial image processing apparatus for extracting facial features and outputting the corresponding coordinate data based on facial image data comprises:
-
facial feature searching range setting means for setting, for each facial feature to be extracted, a searching range based on the coordinate data of specific points and referenced to a particular one of the specific points specified for that facial feature, said specific points being pre-defined in the facial image data for each of the right and left eyes and the mouth; wherein said facial feature searching range setting means comprises mouth searching range setting means and facial contour searching range setting means which sets ranges in the x-axis and y-axis directions to include facial contours referenced to the points specified as the right and left eyes as the searching ranges for the facial contours in the areas at the eye height in the y-axis direction based on the average of the differences between the y-axis coordinate of the point specified as the mouth and the y-axis coordinates of the points specified as the right and left eyes, where the x-coordinate axis represents the side-to-side direction of the facial image and the y-coordinate axis represents the vertical direction of the facial image; and facial feature extracting means for extracting an area of each facial feature, including the mouth, within the searching range set by each corresponding facial feature searching range setting means; wherein it is determined whether the extracted mouth area is correct by determining whether a line connecting right and left end points in the x-axis direction of the mouth area is greater than a predetermined angle relative to a line through a particular point in each of the right and left eyes. - View Dependent Claims (18, 19, 20, 21, 22, 23, 24)
-
-
25. A facial image processing apparatus for extracting facial features and outputting corresponding coordinate data based on facial image data comprises:
-
facial feature searching range setting means for setting, for each facial feature to be extracted, a searching range based on a plurality of specific points including contour points and referenced to a particular one of the plurality of the specific points specified for that facial feature, said specific points being pre-defined in the facial image data for at least the right and left eyes and the mouth; facial feature extracting means for extracting an area of each facial feature, including the mouth, within the searching range set by each corresponding facial feature searching range setting means; wherein said searching range setting means sets ranges for facial contours in areas outside of the eyes and in an area on at least one side of the mouth to include said facial contours referenced to the points specified as the right and left eyes based on the average of the differences between the y-axis coordinate of the point specified as the mouth and y-axis coordinates of the points specified as the right and left eyes, where the x-coordinate axis represents the side-to-side direction of the facial image and the y-coordinate axis represents the vertical direction of the facial image; contour point extracting means for extracting the coordinate data of contour points forming the contour of each facial feature based on the respective areas obtained by the facial feature extracting means; and extracted content evaluating means for evaluating, based on the coordinate values of a specific point in the area of each facial feature, the relative positions of the extracted feature images to determine whether said facial features were correctly extracted, wherein the evaluation process for determining correct extraction of the mouth area includes determining whether a line connecting right and left end points in the x-axis direction of the mouth area is greater than a predetermined angle relative to a line through a particular point in each of the right and left eyes. - View Dependent Claims (27, 28)
-
-
26. A facial image processing apparatus for extracting facial features and outputting the corresponding coordinate data based on facial image data comprising:
-
facial feature searching range setting means for setting, for each facial feature to be extracted, a searching range based on a plurality of specific points including contour points and referenced to a particular one of the plurality of the specific points specified for that facial feature, said specific points being pre-defined in the facial image data for at least the right and left eyes and the mouth; facial feature extracting means for extracting an area of each facial feature, including the mouth, within the searching range set by each corresponding facial feature searching range setting means; contour point extracting means for extracting the coordinate data of contour points forming the contour of each facial feature based on the respective areas obtained by the facial feature extracting means; and extracted content evaluating means for evaluating, based on the coordinate values of a specific point in the area of each facial feature, the relative positions of the extracted feature images to determine whether said facial features were correctly extracted;
whereinthe extracted content evaluating means, in the evaluation process for determining correct extraction of the mouth area extracted by the mouth area extracting means, compares the y-axis coordinates of the right and left end points in the x-axis direction of the extracted mouth area with the y-axis coordinate of the midpoint in the x-axis direction at the top edge of the extracted mouth area where the x-coordinate axis represents the side-to-side direction of the facial image and the y-coordinate axis represents the vertical direction of the facial image; and determines the extracted mouth area to represent an open mouth when the y-axis coordinate of the midpoint in the x-axis direction is smaller than the y-axis coordinate of either right or left end point in the x-axis direction of said mouth area, determines the extracted mouth to be the lower lip only, and sets a searching range for extracting the upper lip.
-
-
29. A facial image processing apparatus for extracting facial features and outputting the corresponding coordinate data based on facial image data comprises:
-
facial feature extracting means for extracting the area of each facial feature using a coordinate system wherein the x-coordinate axis represents the side-to-side direction of the facial image and the y-coordinate axis represents the vertical direction of the facial image; and facial feature contour point extracting means for obtaining the contour points forming the contour of the eyes or mouth in the facial feature area extracted by the facial feature extracting means, and then obtaining from said contour points the coordinates of the right and left end points in the x-axis direction of said facial feature area, then setting, in the area near the midpoint between said right and left end points, a searching range with a width in the x-axis direction equal to a known 1/n fraction of the distance between the right and left end points, and obtaining within said searching range the coordinates of the top and bottom end points in the y-axis direction of the eye or mouth area. - View Dependent Claims (30)
-
-
31. A facial image processing apparatus for extracting facial features and outputting the corresponding coordinate data based on facial image data comprising:
-
facial feature extracting means for extracting the area of each facial feature; contour point extracting means for obtaining the contour points forming the contour of the facial feature based on the area extracted by the facial feature extracting means; and extracted content evaluating means for evaluating the relative positions of the extracted areas based on the coordinate values of a specific point in each facial feature area to determine whether the facial feature areas were correctly extracted;
wherein;the extracted content evaluating means, when the extracted feature is the mouth, determines whether the extracted area is correct by determining whether a line connecting the right and left end points in the x-axis direction of the mouth area is sloped greater than a predetermined angle relative to a line through a specific point in each of the right and left eyes, where the x-coordinate axis represents the side-to-side direction of the facial image and the y-coordinate axis represents the vertical direction of the facial image.
-
-
32. A facial image processing apparatus for extracting facial features and outputting the corresponding coordinate data based on facial image data comprising:
-
facial feature extracting means for extracting the area of each facial feature; contour point extracting means for obtaining the contour points forming the contour of the facial feature based on the area extracted by the facial feature extracting means; and extracted content evaluating means for evaluating the relative positions of the extracted areas based on the coordinate values of a specific point in each facial feature area to determine whether the facial feature areas were correctly extracted;
wherein;the extracted content evaluating means, when the extracted area is a facial contour area, determines whether the extracted area is correct by processing the plural contour points forming the facial contour in units of three adjacent contour points, and comparing the coordinates of the middle of each three contour points with the average of the coordinates of the two end contour points, and substitutes the average of the coordinates of the two end contour points for the coordinates of the middle contour point if the position of the middle contour point is determined to be to the inside of the face relative to the adjacent contour points, where the x-coordinate axis represents the side-to-side direction of the facial image and the y-coordinate axis represents the vertical direction of the facial image.
-
-
33. A computer implemented image processing method, comprising the steps of:
-
scanning and digitizing an image into a first portion of computer memory; identifying a plurality of specific feature points present in the digitized image, wherein each identified specific feature point corresponds to at least one distinct image feature, including right and left eyes and a mouth, present within the digitized image; determining an image searching range for each image feature, including the mouth, facial contours horizontally adjacent the mouth and eyes, based on the identified specific feature points and a predetermined set of feature-classification dependent rules, wherein the image searching range for said facial contours horizontally adjacent the mouth and eyes is determined based on a difference between vertical positions of the specific feature point for each eye and the mouth; extracting feature characteristics corresponding to each image feature, including the mouth, by evaluating digitized image portions defined by the determined source searching ranges; determining whether the extracted mouth is correct by determining whether a line connecting right and left end points in the x-axis direction of the mouth is greater than a predetermined angle relative to a line through a particular point in each of the right and left eyes; and storing the extracted feature characteristics as coordinate data within a second portion of the computer memory. - View Dependent Claims (34, 35, 36, 37, 38, 39, 40, 43, 44)
-
-
41. A computer implemented image processing method, comprising the steps of:
-
scanning and digitizing a facial image into a first portion of computer memory; identifying a plurality of specific feature points present in the digitized facial image, wherein each identified specific feature point corresponds to at least one distinct facial image feature, including at least a mouth, present within the digitized image; determining an image searching range for each facial image feature based on the identified specific feature points and a predetermined set of feature-classification dependent rules; extracting feature characteristics corresponding to each image feature by evaluating digitized image portions defined by the determined source searching ranges; storing the extracted feature characteristics as coordinate data within a second portion of the computer memory; and evaluating the coordinate data for at least one facial feature against the corresponding specific feature point to determine whether the feature was correctly extracted, wherein the coordinate data evaluating step comprises; retrieving coordinate data corresponding to the mouth of the digitized facial image; identifying mouth corner points within the retrieved coordinate data; calculating a lip cusp based on the identified mouth corner points; determining whether the lip cusp is positioned underneath either of the identified mouth corner points; and setting an additional searching range for extracting an upper lip portion of the mouth when it is determined in said lip cusp positioning determining step that the lip cusp is positioned beneath one of the identified mouth corner points.
-
-
42. A computer implemented image processing method, comprising the steps of:
-
scanning and digitizing a facial image into a first portion of computer memory; identifying a plurality of specific feature points present in the digitized facial image, wherein each identified specific feature point corresponds to at least one distinct facial image feature, including at least right and left eyes and a mouth, present within the digitized image; determining an image searching range for each facial image feature, including the mouth, based on the identified specific feature points and a predetermined set of feature-classification dependent rules; extracting feature characteristics corresponding to each image feature, including the mouth, by evaluating digitized image portions defined by the determined source searching ranges; storing the extracted feature characteristics as coordinate data within a second portion of the computer memory; and evaluating the coordinate data for at least one facial feature against the corresponding specific feature point to determine whether the feature was correctly extracted, wherein the coordinate data evaluating step comprises; retrieving coordinate data corresponding to the mouth of the digitized facial image; determining whether the retrieved coordinate data includes the specific feature point for the mouth; and determining whether a line connecting right and left end points in the x-axis direction of the mouth is greater than a predetermined angle relative to a line through a particular point in each of the right and left eyes.
-
-
45. An article of manufacture, comprising a computer usable medium having readable code embodied therein for processing a digitized image, the computer readable program code comprising:
-
computer readable program code for identifying a plurality of specific feature points present in the digitized image, wherein each identified specific feature point corresponds to at least one distinct image feature, including right and left eyes and a mouth, present within the digitized image; computer readable program code for determining an image searching range for each image feature, including a mouth and facial contours horizontally adjacent the mouth and eyes, based on the identified specific feature points and a predetermined set of feature-classification dependent rules, wherein the image range for said facial contours horizontally adjacent the mouth and eyes is determined based on a difference between vertical positions of the specific feature point for each eye and the mouth; computer readable program code for extracting feature characteristics corresponding to each image feature, including the mouth, by evaluating digitized image portions defined by the determined source searching ranges; computer readable program code for determining whether the extracted mouth is correct by determining whether a line connecting right and left end points in the x-axis direction of the mouth is greater than a predetermined angle relative to a line through a particular point in each of the right and left eyes; and computer readable program code for storing the extracted feature characteristics as coordinate data within computer memory. - View Dependent Claims (46, 47, 48, 49, 50, 51, 52, 54, 55, 56)
-
-
53. An article of manufacture, comprising a computer usable medium having readable code embodied therein for processing a digitized image, the computer readable program code comprising:
-
computer readable program code for identifying a plurality of specific feature points present in the digitized image, wherein each identified specific feature point corresponds to at least one distinct image feature, including at least a mouth, present within the digitized image; computer readable program code for determining an image searching range for each image feature based on the identified specific feature points and a predetermined set of feature-classification dependent rules; computer readable program code for extracting feature characteristics corresponding to each image feature by evaluating digitized image portions defined by the determined source searching ranges; computer readable program code for storing the extracted feature characteristics as coordinate data within computer memory; and computer readable code for evaluating the coordinate data for at least one feature against the corresponding specific feature point to determine whether the feature was correctly extracted, wherein said computer readable coordinate data evaluating code comprises; computer readable program code for retrieving coordinate data corresponding to the mouth of the digitized facial image; computer readable program code for identifying mouth corner points within the retrieved coordinate data; computer readable program code for calculating a lip cusp based on the identified mouth corner points; computer readable program code for determining whether the lip cusp is positioned underneath either of the identified mouth corner points; and computer readable program code for setting an additional searching range for extracting an upper lip portion of the mouth when it is determined by said lip cusp positioning determining means that the lip cusp is positioned beneath one of the identified mouth corner points.
-
Specification