System and method for effectively extracting facial feature information

Claim 1

A system for effectively analyzing image data, comprising: a facial feature detector configured to generate facial feature information by analyzing said image data, said facial feature information including eye-area information and mouth-area information, said facial feature detector performing an eye-area vertical projection procedure for generating said eye-area information, said facial feature detector also separately performing a mouth-area vertical projection procedure for generating said mouth-area information, said facial feature detector further utilizing a feature filter to refine said facial feature information for a facial feature, said feature filter having a plurality of segments that collectively are shaped to represent said facial feature, said facial feature detector moving said feature filter over pixels of said image data to repeatedly calculate feature merit values from said pixels currently aligned with said feature filter, said feature merit values being determined by calculating sums of pixel values corresponding with each of said segments and then calculating a weighted difference of said sums, an optimal one of said feature merit values accurately locating said facial feature in said image data; and a processor configured to control said facial feature detector to thereby effectively analyze said image data.

Claim 2

A system for effectively analyzing image data, comprising: a facial feature detector configured to generate facial feature information by analyzing said image data, said facial feature information including eye-area information and mouth-area information, said facial feature detector performing an eye-area vertical projection procedure for generating said eye-area information, said facial feature detector also separately performing a mouth-area vertical projection procedure for generating said mouth-area information, said facial feature detector further utilizing a feature filter to refine said facial feature information for a facial feature, said feature filter having a plurality of segments that collectively are shaped to represent said facial feature, said facial feature detector moving said feature filter over pixels of said image data to repeatedly calculate feature merit values from said pixels currently aligned with said feature filter, said feature merit values being determined by calculating sums of pixel values corresponding with each of said segments and then calculating a weighted difference of said sums, an optimal one of said feature merit values accurately locating said facial feature in said image data; and a processor configured to control said facial feature detector to thereby effectively analyze said image data.

Claim 17

A system for effectively analyzing image data, comprising: a facial feature detector configured to generate facial feature information by analyzing said image data, said facial feature information including eye-area information and mouth-area information, said facial feature detector performing an eye-area vertical projection procedure for generating said eye-area information, said facial feature detector also separately performing a mouth-area vertical projection procedure for generating said mouth-area information, said facial feature detector further utilizing a feature filter to refine said facial feature information for a facial feature, said feature filter having a plurality of segments that collectively are shaped to represent said facial feature, said facial feature detector moving said feature filter over pixels of said image data to repeatedly calculate feature merit values from said pixels currently aligned with said feature filter, an optimal one of said feature merit values accurately locating said facial feature in said image data, a face detector module performing an initial face detection procedure to convert said image data into a gray-scale format, said face detector module then generating approximate location information for a subject face in said image data, said facial feature detector performing a face rotation procedure upon said subject face in said image data to realign said subject face in a vertical alignment, said facial feature detector creating an eye-area window around an approximate eye area in said subject face, said eye-area window including an eye and an eyebrow, said facial feature detector performing said eye-area vertical projection procedure by summing gray-scale values of pixels in each horizontal line of said eye-area window to produce a corresponding histogram which represents results of said eye-area vertical projection procedure, said facial feature detector identifying a first low-amplitude point in said histogram as corresponding to pixels of an eyebrow region in said subject face, said facial feature detector also identifying a second low-amplitude point in said histogram as corresponding to pixels of an eye region in said subject face, said facial feature detector determining a correct separator area between said first low-amplitude point and said second low-amplitude point by selecting a maximum separator merit value of separator area candidates according to a formula: Merit Value=Separator Candidate Value−10 (Eyebrow Value−Eye Value); and a processor configured to control said facial feature detector to thereby effectively analyze said image data.

Claim 18

A system for effectively analyzing image data, comprising: a facial feature detector configured to generate facial feature information by analyzing said image data, said facial feature information including eye-area information and mouth-area information, said facial feature detector performing an eye-area vertical projection procedure for generating said eye-area information, said facial feature detector also separately performing a mouth-area vertical projection procedure for generating said mouth-area information, said facial feature detector further utilizing a feature filter to refine said facial feature information for a facial feature, said feature filter having a plurality of segments that collectively are shaped to represent said facial feature, said facial feature detector moving said feature filter over pixels of said image data to repeatedly calculate feature merit values from said pixels currently aligned with said feature filter, an optimal one of said feature merit values accurately locating said facial feature in said image data, a face detector module performing an initial face detection procedure to convert said image data into a gray-scale format, said face detector module then generating approximate location information for a subject face in said image data, said facial feature detector performing a face rotation procedure upon said subject face in said image data to realign said subject face in a vertical alignment, said facial feature detector creating an eye-area window around an approximate eye area in said subject face, said eye-area window including an eye and an eyebrow, said facial feature detector performing said eye-area vertical projection procedure by summing gray-scale values of pixels in each horizontal line of said eye-area window to produce a corresponding histogram which represents results of said eye-area vertical projection procedure, said facial feature detector identifying a first low-amplitude point in said histogram as corresponding to pixels of an eyebrow region in said subject face, said facial feature detector also identifying a second low-amplitude point in said histogram as corresponding to pixels of an eye region in said subject face, said facial feature detector determining a correct separator area between said first low-amplitude point and said second low-amplitude point by selecting a maximum separator merit value of separator area candidates according to a formula: Merit Value=Separator Candidate Value−10 (Eyebrow Value−Eye Value); and a processor configured to control said facial feature detector to thereby effectively analyze said image data.

Claim 19

A system for effectively analyzing image data, comprising: a facial feature detector configured to generate facial feature information by analyzing said image data, said facial feature information including eye-area information and mouth-area information, said facial feature detector performing an eye-area vertical projection procedure for generating said eye-area information, said facial feature detector also performing a mouth-area vertical projection procedure for generating said mouth-area information; a face detector module that performs an initial face detection procedure to convert said image data into a gray-scale format, said face detector module then generating approximate location information for a subject face in said image data, said facial feature detector performing a face rotation procedure upon said subject face in said image data to realign said subject face in a vertical alignment, said facial feature detector creating an eye-area window around an approximate eye area in said subject face, said eye-area window including an eye and an eyebrow, said facial feature detector performing said eye-area vertical projection procedure by summing gray-scale values of pixels in each horizontal line of said eye-area window to produce a corresponding histogram which represents results of said eye-area vertical projection procedure, said facial feature detector identifying a first low-amplitude point in said histogram as corresponding to pixels of an eyebrow region in said subject face, said facial feature detector also identifying a second low-amplitude point in said histogram as corresponding to pixels of an eye region in said subject face, said facial feature detector utilizing an eyebrow filter to detect an upper edge and a slope angle of said eyebrow from said subject face, said eyebrow filter being implemented to have a filter width of “M” and a filter height of “M/4”, where “M” is an approximate distance between two eyes of said subject face, said facial feature detector moving said eyebrow filter, pixel by pixel, to different locations of said eyebrow region in said eye-area window, said facial feature detector, at each of said different locations, adding all pixel values in an upper portion of said eyebrow filter to produce an upper value, said facial feature detector also adding all pixel values in a lower portion of said eyebrow filter to produce a lower value, said facial feature detector then subtracting said lower value from said upper value to produce a local eyebrow value for each corresponding position of said eyebrow filter, said facial feature detector then selecting a maximum local eyebrow value to accurately detect a position and a slope angle of a corresponding eyebrow; and a processor configured to control said facial feature detector to thereby effectively analyze said image data.

Claim 20

A system for effectively analyzing image data, comprising: a facial feature detector configured to generate facial feature information by analyzing said image data, said facial feature information including eye-area information and mouth-area information, said facial feature detector performing an eye-area vertical projection procedure for generating said eye-area information, said facial feature detector also performing a mouth-area vertical projection procedure for generating said mouth-area information; a face detector module that performs an initial face detection procedure to convert said image data into a gray-scale format, said face detector module then generating approximate location information for a subject face in said image data, said facial feature detector performing a face rotation procedure upon said subject face in said image data to realign said subject face in a vertical alignment, said facial feature detector creating an eye-area window around an approximate eye area in said subject face, said eye-area window including an eye and an eyebrow, said facial feature detector performing said eye-area vertical projection procedure by summing gray-scale values of pixels in each horizontal line of said eye-area window to produce a corresponding histogram which represents results of said eye-area vertical projection procedure, said facial feature detector identifying a first low-amplitude point in said histogram as corresponding to pixels of an eyebrow region in said subject face, said facial feature detector also identifying a second low-amplitude point in said histogram as corresponding to pixels of an eye region in said subject face, said facial feature detector utilizing an eyebrow filter to detect an upper edge and a slope angle of said eyebrow from said subject face, said eyebrow filter being implemented to have a filter width of “M” and a filter height of “M/4”, where “M” is an approximate distance between two eyes of said subject face, said facial feature detector moving said eyebrow filter, pixel by pixel, to different locations of said eyebrow region in said eye-area window, said facial feature detector, at each of said different locations, adding all pixel values in an upper portion of said eyebrow filter to produce an upper value, said facial feature detector also adding all pixel values in a lower portion of said eyebrow filter to produce a lower value, said facial feature detector then subtracting said lower value from said upper value to produce a local eyebrow value for each corresponding position of said eyebrow filter, said facial feature detector then selecting a maximum local eyebrow value to accurately detect a position and a slope angle of a corresponding eyebrow; and a processor configured to control said facial feature detector to thereby effectively analyze said image data.

Claim 22

A system for effectively analyzing image data, comprising: a facial feature detector configured to generate facial feature information by analyzing said image data, said facial feature information including eye-area information and mouth-area information, said facial feature detector performing an eye-area vertical projection procedure for generating said eye-area information, said facial feature detector also performing a mouth-area vertical projection procedure for generating said mouth-area information; a face detector module that performs an initial face detection procedure to convert said image data into a gray-scale format, said face detector module then generating approximate location information for a subject face in said image data, said facial feature detector performing a face rotation procedure upon said subject face in said image data to realign said subject face in a vertical alignment, said facial feature detector creating an eye-area window around an approximate eye area in said subject face, said eye-area window including an eye and an eyebrow, said facial feature detector performing said eye-area vertical projection procedure by summing gray-scale values of pixels in each horizontal line of said eye-area window to produce a corresponding histogram which represents results of said eye-area vertical projection procedure, said facial feature detector identifying a first low-amplitude point in said histogram as corresponding to pixels of an eyebrow region in said subject face, said facial feature detector also identifying a second low-amplitude point in said histogram as corresponding to pixels of an eye region in said subject face, said facial feature detector utilizing an iris filter to detect coordinates for an iris region in said eye-area of said subject face, said iris filter being implemented to include an iris area with an iris radius “r” that is approximately equal to 0.09דM”, where “M” is an approximate distance between eyes in said subject face, said iris filter also being implemented with a pupil area having a pupil radius that is approximately equal to “r/2”, where “r” is said iris radius, said iris filter further being implemented with a eye-white area having an eye-white radius that is approximately equal to “1.5×r”, where “r” is said iris radius; and a processor configured to control said facial feature detector to thereby effectively analyze said image data.

Claim 23

A system for effectively analyzing image data, comprising: a facial feature detector configured to generate facial feature information by analyzing said image data, said facial feature information including eye-area information and mouth-area information, said facial feature detector performing an eye-area vertical projection procedure for generating said eye-area information, said facial feature detector also performing a mouth-area vertical projection procedure for generating said mouth-area information; a face detector module that performs an initial face detection procedure to convert said image data into a gray-scale format, said face detector module then generating approximate location information for a subject face in said image data, said facial feature detector performing a face rotation procedure upon said subject face in said image data to realign said subject face in a vertical alignment, said facial feature detector creating an eye-area window around an approximate eye area in said subject face, said eye-area window including an eye and an eyebrow, said facial feature detector performing said eye-area vertical projection procedure by summing gray-scale values of pixels in each horizontal line of said eye-area window to produce a corresponding histogram which represents results of said eye-area vertical projection procedure, said facial feature detector identifying a first low-amplitude point in said histogram as corresponding to pixels of an eyebrow region in said subject face, said facial feature detector also identifying a second low-amplitude point in said histogram as corresponding to pixels of an eye region in said subject face, said facial feature detector utilizing an iris filter to detect coordinates for an iris region in said eye-area of said subject face, said iris filter being implemented to include an iris area with an iris radius “r” that is approximately equal to 0.09דM”, where “M” is an approximate distance between eyes in said subject face, said iris filter also being implemented with a pupil area having a pupil radius that is approximately equal to “r/2”, where “r” is said iris radius, said iris filter further being implemented with a eye-white area having an eye-white radius that is approximately equal to “1.5×r”, where “r” is said iris radius; and a processor configured to control said facial feature detector to thereby effectively analyze said image data.

Claim 24

A system for effectively analyzing image data, comprising: a facial feature detector configured to generate facial feature information by analyzing said image data, said facial feature information including eye-area information and mouth-area information, said facial feature detector performing an eye-area vertical projection procedure for generating said eye-area information, said facial feature detector also performing a mouth-area vertical projection procedure for generating said mouth-area information; a face detector module that performs an initial face detection procedure to convert said image data into a gray-scale format, said face detector module then generating approximate location information for a subject face in said image data, said facial feature detector performing a face rotation procedure upon said subject face in said image data to realign said subject face in a vertical alignment, said facial feature detector creating an eye-area window around an approximate eye area in said subject face, said eye-area window including an eye and an eyebrow, said facial feature detector performing said eye-area vertical projection procedure by summing gray-scale values of pixels in each horizontal line of said eye-area window to produce a corresponding histogram which represents results of said eye-area vertical projection procedure, said facial feature detector identifying a first low-amplitude point in said histogram as corresponding to pixels of an eyebrow region in said subject face, said facial feature detector also identifying a second low-amplitude point in said histogram as corresponding to pixels of an eye region in said subject face, said facial feature detector utilizing an iris filter to detect coordinates for an iris region in said eye-area of said subject face, said facial feature detector moving said iris filter, pixel by pixel, to different locations in an eye-area window around said eye area, said facial feature detector, at each of said different locations, generating a white-area value that is equal to a sum all pixel values in said eye-white area of said iris filter, said facial feature detector also generating an eyelid value that is equal to a sum all pixel values in an eyelid area of said iris filter, said facial feature detector further generating an iris value that is equal to a sum of all pixel values in said iris area, said facial feature detector then calculating an iris merit value at each of said different locations according to a formula: Iris Merit Value=(2×white-area value)−(3×eyelid value)−(6×iris value) said facial feature detector then selecting a maximum iris merit value to accurately detect coordinates of said iris area; and a processor configured to control said facial feature detector to thereby effectively analyze said image data.

Claim 25

A system for effectively analyzing image data, comprising: a facial feature detector configured to generate facial feature information by analyzing said image data, said facial feature information including eye-area information and mouth-area information, said facial feature detector performing an eye-area vertical projection procedure for generating said eye-area information, said facial feature detector also performing a mouth-area vertical projection procedure for generating said mouth-area information; a face detector module that performs an initial face detection procedure to convert said image data into a gray-scale format, said face detector module then generating approximate location information for a subject face in said image data, said facial feature detector performing a face rotation procedure upon said subject face in said image data to realign said subject face in a vertical alignment, said facial feature detector creating an eye-area window around an approximate eye area in said subject face, said eye-area window including an eye and an eyebrow, said facial feature detector performing said eye-area vertical projection procedure by summing gray-scale values of pixels in each horizontal line of said eye-area window to produce a corresponding histogram which represents results of said eye-area vertical projection procedure, said facial feature detector identifying a first low-amplitude point in said histogram as corresponding to pixels of an eyebrow region in said subject face, said facial feature detector also identifying a second low-amplitude point in said histogram as corresponding to pixels of an eye region in said subject face, said facial feature detector utilizing an iris filter to detect coordinates for an iris region in said eye-area of said subject face, said facial feature detector moving said iris filter, pixel by pixel, to different locations in an eye-area window around said eye area, said facial feature detector, at each of said different locations, generating a white-area value that is equal to a sum all pixel values in said eye-white area of said iris filter, said facial feature detector also generating an eyelid value that is equal to a sum all pixel values in an eyelid area of said iris filter, said facial feature detector further generating an iris value that is equal to a sum of all pixel values in said iris area, said facial feature detector then calculating an iris merit value at each of said different locations according to a formula: Iris Merit Value=(2×white-area value)−(3×eyelid value)−(6×iris value) said facial feature detector then selecting a maximum iris merit value to accurately detect coordinates of said iris area; and a processor configured to control said facial feature detector to thereby effectively analyze said image data.

Claim 26

A method for effectively analyzing image data, comprising the steps of: analyzing said image data with a facial feature detector to generate facial feature information, said facial feature information including eye-area information and mouth-area information, said facial feature detector performing an eye-area vertical projection procedure for generating said eye-area information, said facial feature detector also separately performing a mouth-area vertical projection procedure for generating said mouth-area information; utilizing a feature filter to refine said facial feature information for a facial feature, said feature filter having a plurality of segments that collectively are shaped to represent said facial feature, said facial feature detector moving said feature filter over pixels of said image data to repeatedly calculate feature merit values from said pixels currently aligned with said feature filter, said feature merit values being determined by calculating sums of pixel values corresponding with each of said segments and then calculating a weighted difference of said sums, an optimal one of said feature merit values accurately locating said facial feature in said image data; and controlling said facial feature detector with a processor to thereby effectively analyze said image data.

Claim 27

A method for effectively analyzing image data, comprising the steps of: analyzing said image data with a facial feature detector to generate facial feature information, said facial feature information including eye-area information and mouth-area information, said facial feature detector performing an eye-area vertical projection procedure for generating said eye-area information, said facial feature detector also separately performing a mouth-area vertical projection procedure for generating said mouth-area information; utilizing a feature filter to refine said facial feature information for a facial feature, said feature filter having a plurality of segments that collectively are shaped to represent said facial feature, said facial feature detector moving said feature filter over pixels of said image data to repeatedly calculate feature merit values from said pixels currently aligned with said feature filter, said feature merit values being determined by calculating sums of pixel values corresponding with each of said segments and then calculating a weighted difference of said sums, an optimal one of said feature merit values accurately locating said facial feature in said image data; and controlling said facial feature detector with a processor to thereby effectively analyze said image data.

Claim 42

A method for effectively analyzing image data, comprising the steps of: analyzing said image data with a facial feature detector to generate facial feature information, said facial feature information including eye-area information and mouth-area information, said facial feature detector performing an eye-area vertical projection procedure for generating said eye-area information, said facial feature detector also separately performing a mouth-area vertical projection procedure for generating said mouth-area information; utilizing a feature filter to refine said facial feature information for a facial feature, said feature filter having a plurality of segments that collectively are shaped to represent said facial feature, said facial feature detector moving said feature filter over pixels of said image data to repeatedly calculate feature merit values from said pixels currently aligned with said feature filter, an optimal one of said feature merit values accurately locating said facial feature in said image data; a face detector module performing an initial face detection procedure to convert said image data into a gray-scale format, said face detector module then generating approximate location information for a subject face in said image data, said facial feature detector performing a face rotation procedure upon said subject face in said image data to realign said subject face in a vertical alignment, said facial feature detector creating an eye-area window around an approximate eye area in said subject face, said eye-area window including an eye and an eyebrow, said facial feature detector performing said eye-area vertical projection procedure by summing gray-scale values of pixels in each horizontal line of said eye-area window to produce a corresponding histogram which represents results of said eye-area vertical projection procedure, said facial feature detector identifying a first low-amplitude point in said histogram as corresponding to pixels of an eyebrow region in said subject face, said facial feature detector also identifying a second low-amplitude point in said histogram as corresponding to pixels of an eye region in said subject face, said facial feature detector determining a correct separator area between said first low-amplitude point and said second low-amplitude point by selecting a maximum separator merit value of separator area candidates according to a formula: Merit Value=Separator Candidate Value−10 (Eyebrow Value−Eye Value); and controlling said facial feature detector with a processor to thereby effectively analyze said image data.

Claim 43

A method for effectively analyzing image data, comprising the steps of: analyzing said image data with a facial feature detector to generate facial feature information, said facial feature information including eye-area information and mouth-area information, said facial feature detector performing an eye-area vertical projection procedure for generating said eye-area information, said facial feature detector also separately performing a mouth-area vertical projection procedure for generating said mouth-area information; utilizing a feature filter to refine said facial feature information for a facial feature, said feature filter having a plurality of segments that collectively are shaped to represent said facial feature, said facial feature detector moving said feature filter over pixels of said image data to repeatedly calculate feature merit values from said pixels currently aligned with said feature filter, an optimal one of said feature merit values accurately locating said facial feature in said image data; a face detector module performing an initial face detection procedure to convert said image data into a gray-scale format, said face detector module then generating approximate location information for a subject face in said image data, said facial feature detector performing a face rotation procedure upon said subject face in said image data to realign said subject face in a vertical alignment, said facial feature detector creating an eye-area window around an approximate eye area in said subject face, said eye-area window including an eye and an eyebrow, said facial feature detector performing said eye-area vertical projection procedure by summing gray-scale values of pixels in each horizontal line of said eye-area window to produce a corresponding histogram which represents results of said eye-area vertical projection procedure, said facial feature detector identifying a first low-amplitude point in said histogram as corresponding to pixels of an eyebrow region in said subject face, said facial feature detector also identifying a second low-amplitude point in said histogram as corresponding to pixels of an eye region in said subject face, said facial feature detector determining a correct separator area between said first low-amplitude point and said second low-amplitude point by selecting a maximum separator merit value of separator area candidates according to a formula: Merit Value=Separator Candidate Value−10 (Eyebrow Value−Eye Value); and controlling said facial feature detector with a processor to thereby effectively analyze said image data.

Claim 44

A method for effectively analyzing image data, comprising the steps of: analyzing said image data with a facial feature detector to generate facial feature information, said facial feature information including eye-area information and mouth-area information, said facial feature detector performing an eye-area vertical projection procedure for generating said eye-area information, said facial feature detector also performing a mouth-area vertical projection procedure for generating said mouth-area information; utilizing a face detector module to perform an initial face detection procedure to convert said image data into a gray-scale format, said face detector module then generating approximate location information for a subject face in said image data, said facial feature detector performing a face rotation procedure upon said subject face in said image data to realign said subject face in a vertical alignment, said facial feature detector creating an eye-area window around an approximate eye area in said subject face, said eye-area window including an eye and an eyebrow, said facial feature detector performing said eye-area vertical projection procedure by summing gray-scale values of pixels in each horizontal line of said eye-area window to produce a corresponding histogram which represents results of said eye-area vertical projection procedure, said facial feature detector identifying a first low-amplitude point in said histogram as corresponding to pixels of an eyebrow region in said subject face, said facial feature detector also identifying a second low-amplitude point in said histogram as corresponding to pixels of an eye region in said subject face, said facial feature detector utilizing an eyebrow filter to detect an upper edge and a slope angle of said eyebrow from said subject face, said eyebrow filter being implemented to have a filter width of “M” and a filter height of “M/4”, where “M” is an approximate distance between two eyes of said subject face, said facial feature detector moving said eyebrow filter, pixel by pixel, to different locations of said eyebrow region in said eye-area window, said facial feature detector, at each of said different locations, adding all pixel values in an upper portion of said eyebrow filter to produce an upper value, said facial feature detector also adding all pixel values in a lower portion of said eyebrow filter to produce a lower value, said facial feature detector then subtracting said lower value from said upper value to produce a local eyebrow value for each corresponding position of said eyebrow filter, said facial feature detector then selecting a maximum local eyebrow value to accurately detect a position and a slope angle of a corresponding eyebrow; and controlling said facial feature detector with a processor to thereby effectively analyze said image.

Claim 45

A method for effectively analyzing image data, comprising the steps of: analyzing said image data with a facial feature detector to generate facial feature information, said facial feature information including eye-area information and mouth-area information, said facial feature detector performing an eye-area vertical projection procedure for generating said eye-area information, said facial feature detector also performing a mouth-area vertical projection procedure for generating said mouth-area information; utilizing a face detector module to perform an initial face detection procedure to convert said image data into a gray-scale format, said face detector module then generating approximate location information for a subject face in said image data, said facial feature detector performing a face rotation procedure upon said subject face in said image data to realign said subject face in a vertical alignment, said facial feature detector creating an eye-area window around an approximate eye area in said subject face, said eye-area window including an eye and an eyebrow, said facial feature detector performing said eye-area vertical projection procedure by summing gray-scale values of pixels in each horizontal line of said eye-area window to produce a corresponding histogram which represents results of said eye-area vertical projection procedure, said facial feature detector identifying a first low-amplitude point in said histogram as corresponding to pixels of an eyebrow region in said subject face, said facial feature detector also identifying a second low-amplitude point in said histogram as corresponding to pixels of an eye region in said subject face, said facial feature detector utilizing an eyebrow filter to detect an upper edge and a slope angle of said eyebrow from said subject face, said eyebrow filter being implemented to have a filter width of “M” and a filter height of “M/4”, where “M” is an approximate distance between two eyes of said subject face, said facial feature detector moving said eyebrow filter, pixel by pixel, to different locations of said eyebrow region in said eye-area window, said facial feature detector, at each of said different locations, adding all pixel values in an upper portion of said eyebrow filter to produce an upper value, said facial feature detector also adding all pixel values in a lower portion of said eyebrow filter to produce a lower value, said facial feature detector then subtracting said lower value from said upper value to produce a local eyebrow value for each corresponding position of said eyebrow filter, said facial feature detector then selecting a maximum local eyebrow value to accurately detect a position and a slope angle of a corresponding eyebrow; and controlling said facial feature detector with a processor to thereby effectively analyze said image.

Claim 47

A method for effectively analyzing image data, comprising the steps of: analyzing said image data with a facial feature detector to generate facial feature information, said facial feature information including eye-area information and mouth-area information, said facial feature detector performing an eye-area vertical projection procedure for generating said eye-area information, said facial feature detector also performing a mouth-area vertical projection procedure for generating said mouth-area information; utilizing a face detector module to perform an initial face detection procedure to convert said image data into a gray-scale format, said face detector module then generating approximate location information for a subject face in said image data, said facial feature detector performing a face rotation procedure upon said subject face in said image data to realign said subject face in a vertical alignment, said facial feature detector creating an eye-area window around an approximate eye area in said subject face, said eye-area window including an eye and an eyebrow, said facial feature detector performing said eye-area vertical projection procedure by summing gray-scale values of pixels in each horizontal line of said eye-area window to produce a corresponding histogram which represents results of said eye-area vertical projection procedure, said facial feature detector identifying a first low-amplitude point in said histogram as corresponding to pixels of an eyebrow region in said subject face, said facial feature detector also identifying a second low-amplitude point in said histogram as corresponding to pixels of an eye region in said subject face, said facial feature detector utilizing an iris filter to detect coordinates for an iris region in said eye-area of said subject face, said iris filter being implemented to include an iris area with an iris radius “r” that is approximately equal to 0.09דM”, where “M” is an approximate distance between eyes in said subject face, said iris filter also being implemented with a pupil area having a pupil radius that is approximately equal to “r/2”, where “r” is said iris radius, said iris filter further being implemented with a eye-white area having an eye-white radius that is approximately equal to “1.5×r”, where “r” is said iris radius; and controlling said facial feature detector with a processor to thereby effectively analyze said image data.

Claim 48

A method for effectively analyzing image data, comprising the steps of: analyzing said image data with a facial feature detector to generate facial feature information, said facial feature information including eye-area information and mouth-area information, said facial feature detector performing an eye-area vertical projection procedure for generating said eye-area information, said facial feature detector also performing a mouth-area vertical projection procedure for generating said mouth-area information; utilizing a face detector module to perform an initial face detection procedure to convert said image data into a gray-scale format, said face detector module then generating approximate location information for a subject face in said image data, said facial feature detector performing a face rotation procedure upon said subject face in said image data to realign said subject face in a vertical alignment, said facial feature detector creating an eye-area window around an approximate eye area in said subject face, said eye-area window including an eye and an eyebrow, said facial feature detector performing said eye-area vertical projection procedure by summing gray-scale values of pixels in each horizontal line of said eye-area window to produce a corresponding histogram which represents results of said eye-area vertical projection procedure, said facial feature detector identifying a first low-amplitude point in said histogram as corresponding to pixels of an eyebrow region in said subject face, said facial feature detector also identifying a second low-amplitude point in said histogram as corresponding to pixels of an eye region in said subject face, said facial feature detector utilizing an iris filter to detect coordinates for an iris region in said eye-area of said subject face, said iris filter being implemented to include an iris area with an iris radius “r” that is approximately equal to 0.09דM”, where “M” is an approximate distance between eyes in said subject face, said iris filter also being implemented with a pupil area having a pupil radius that is approximately equal to “r/2”, where “r” is said iris radius, said iris filter further being implemented with a eye-white area having an eye-white radius that is approximately equal to “1.5×r”, where “r” is said iris radius; and controlling said facial feature detector with a processor to thereby effectively analyze said image data.

Claim 49

A method for effectively analyzing image data, comprising the steps of: analyzing said image data with a facial feature detector to generate facial feature information, said facial feature information including eye-area information and mouth-area information, said facial feature detector performing an eye-area vertical projection procedure for generating said eye-area information, said facial feature detector also performing a mouth-area vertical projection procedure for generating said mouth-area information; utilizing a face detector module to perform an initial face detection procedure to convert said image data into a gray-scale format, said face detector module then generating approximate location information for a subject face in said image data, said facial feature detector performing a face rotation procedure upon said subject face in said image data to realign said subject face in a vertical alignment, said facial feature detector creating an eye-area window around an approximate eye area in said subject face, said eye-area window including an eye and an eyebrow, said facial feature detector performing said eye-area vertical projection procedure by summing gray-scale values of pixels in each horizontal line of said eye-area window to produce a corresponding histogram which represents results of said eye-area vertical projection procedure, said facial feature detector identifying a first low-amplitude point in said histogram as corresponding to pixels of an eyebrow region in said subject face, said facial feature detector also identifying a second low-amplitude point in said histogram as corresponding to pixels of an eye region in said subject face, said facial feature detector utilizing an iris filter to detect coordinates for an iris region in said eye-area of said subject face, said facial feature detector moving said iris filter, pixel by pixel, to different locations in an eye-area window around said eye area, said facial feature detector, at each of said different locations, generating a white-area value that is equal to a sum all pixel values in said eye-white area of said iris filter, said facial feature detector also generating an eyelid value that is equal to a sum all pixel values in an eyelid area of said iris filter, said facial feature detector further generating an iris value that is equal to a sum of all pixel values in said iris area, said facial feature detector then calculating an iris merit value at each of said different locations according to a formula: Iris Merit Value=(2×white-area value)−(3×eyelid value)−(6×iris value) said facial feature detector then selecting a maximum iris merit value to accurately detect coordinates of said iris area; and controlling said facial feature detector with a processor to thereby effectively analyze said image data.

Claim 50

A method for effectively analyzing image data, comprising the steps of: analyzing said image data with a facial feature detector to generate facial feature information, said facial feature information including eye-area information and mouth-area information, said facial feature detector performing an eye-area vertical projection procedure for generating said eye-area information, said facial feature detector also performing a mouth-area vertical projection procedure for generating said mouth-area information; utilizing a face detector module to perform an initial face detection procedure to convert said image data into a gray-scale format, said face detector module then generating approximate location information for a subject face in said image data, said facial feature detector performing a face rotation procedure upon said subject face in said image data to realign said subject face in a vertical alignment, said facial feature detector creating an eye-area window around an approximate eye area in said subject face, said eye-area window including an eye and an eyebrow, said facial feature detector performing said eye-area vertical projection procedure by summing gray-scale values of pixels in each horizontal line of said eye-area window to produce a corresponding histogram which represents results of said eye-area vertical projection procedure, said facial feature detector identifying a first low-amplitude point in said histogram as corresponding to pixels of an eyebrow region in said subject face, said facial feature detector also identifying a second low-amplitude point in said histogram as corresponding to pixels of an eye region in said subject face, said facial feature detector utilizing an iris filter to detect coordinates for an iris region in said eye-area of said subject face, said facial feature detector moving said iris filter, pixel by pixel, to different locations in an eye-area window around said eye area, said facial feature detector, at each of said different locations, generating a white-area value that is equal to a sum all pixel values in said eye-white area of said iris filter, said facial feature detector also generating an eyelid value that is equal to a sum all pixel values in an eyelid area of said iris filter, said facial feature detector further generating an iris value that is equal to a sum of all pixel values in said iris area, said facial feature detector then calculating an iris merit value at each of said different locations according to a formula: Iris Merit Value=(2×white-area value)−(3×eyelid value)−(6×iris value) said facial feature detector then selecting a maximum iris merit value to accurately detect coordinates of said iris area; and controlling said facial feature detector with a processor to thereby effectively analyze said image data.

Claim 53

A system for effectively analyzing image data, comprising: means for analyzing said image data to generate facial feature information, said facial feature information said facial feature information including eye-area information and mouth-area information, said means for analyzing performing an eye-area vertical projection procedure for generating said eye-area information, said means for analyzing also performing a mouth-area vertical projection procedure for generating said mouth-area information; means for refining said facial feature information for a facial feature, said means for refining having a plurality of segments that collectively are shaped to represent said facial feature, said means for analyzing moving said means for refining over pixels of said image data to repeatedly calculate feature merit values from said pixels currently aligned with said means for refining, said feature merit values being determined by calculating sums of pixel values corresponding with each of said segments and then calculating a weighted difference of said sums, an optimal one of said feature merit values accurately locating said facial feature in said image data; and means for controlling said facial feature detector to thereby effectively analyze said image data.

Claim 54

A system for effectively analyzing image data, comprising: means for analyzing said image data to generate facial feature information, said facial feature information said facial feature information including eye-area information and mouth-area information, said means for analyzing performing an eye-area vertical projection procedure for generating said eye-area information, said means for analyzing also performing a mouth-area vertical projection procedure for generating said mouth-area information; means for refining said facial feature information for a facial feature, said means for refining having a plurality of segments that collectively are shaped to represent said facial feature, said means for analyzing moving said means for refining over pixels of said image data to repeatedly calculate feature merit values from said pixels currently aligned with said means for refining, said feature merit values being determined by calculating sums of pixel values corresponding with each of said segments and then calculating a weighted difference of said sums, an optimal one of said feature merit values accurately locating said facial feature in said image data; and means for controlling said facial feature detector to thereby effectively analyze said image data.