Method for generating a customized/personalized head related transfer function

Abstract

There is provided a method for generating a personalized Head Related Transfer Function (HRTF). The method can include capturing an image of an ear using a portable device, auto-scaling the captured image to determine physical geometries of the ear and obtaining a personalized HRTF based on the determined physical geometries of the ear.

1 . A method for enhancing audio rendering by generating a customized Head Related Transfer Function (HRTF), the method comprising:
 acquiring a captured image of at least one ear of an individual using an image capturing device configured for processing a preliminary image that is a preliminary version of the captured image to provide feedback to a user;   generating a set of landmarks that correspond to control points from at least the preliminary image by applying a model to the preliminary image;   extracting image based properties from the generated set of landmarks for the individual from a finalized representation of the captured image; and   providing the image based properties to a selection processor configured to select a customized HRTF dataset from a plurality of HRTF datasets that have been determined for a plurality of individuals.   
     
     
         2 . The method as recited in  claim 1 , further comprising:
 scaling the set of landmarks to correspond to the finalized representation of the captured image and wherein the image capturing device is configured for displaying at least the preliminary version of the captured image on a display screen.   
     
     
         3 . The method as recited in  claim 2 , wherein acquiring a captured image further includes generating a landmark template that is a visual guide overlaid in real-time onto the display screen of the image capturing device which provides an indication of alignment to the user when the set of landmarks matched against the preliminary image are within an acceptable tolerance. 
     
     
         4 . The method as recited in  claim 2 , wherein the scaling is performed using from the captured image at least one of a conventionally sized reference object; focal distance data and focal length associated with the focusing of the input image; or a tragus length for the ear in the input image. 
     
     
         5 . The method as recited in  claim 1 , wherein acquiring a captured image includes a detecting step which combines both a determination of the presence of an ear and generation of landmarks. 
     
     
         6 . The method as recited in  claim 1 , wherein the model used in extracting image based properties for the captured image is an Active Shape Model and the Active Shape Model was previously trained on at least a plurality of ear images of individuals. 
     
     
         7 . The method as recited in  claim 1 , wherein the preliminary image is one of several preliminary images iteratively processed by deforming the model to match the preliminary image of the ear. 
     
     
         8 . The method as recited in  claim 1 , wherein the plurality of HRTF datasets is a database having a collection of HRTF pairs at different azimuth and elevation values for each HRTF dataset. 
     
     
         9 . The method as recited in  claim 1 , wherein the customized HRTF dataset from the plurality of HRTF datasets is selected based on matching most closely the extracted image based properties to the corresponding image based properties associated with each of the HRTF datasets in the plurality. 
     
     
         10 . The method as recited in  claim 1 , wherein multiple HRTF datasets are selected from the plurality of HRTF datasets by matching one or more of image based properties and the customized HRTF is generated by one of interpolation or perturbation relating to at least one of the multiple HRTF datasets, and wherein the image based properties comprise one or more of landmark distances, landmark arcs, landmark angles, landmark geometric relationships, a concha length, ear width, ear height, general physical dimensions of the ear, and a three dimensional representation of the ear. 
     
     
         11 . A processing device for processing customized HRTFs comprising:
 an image processing device configured for:
 acquiring a captured image of at least one ear of an individual and processing at least one preliminary image that is a preliminary version of the captured image to provide feedback to a user; 
 generating a set of landmarks that correspond to control points by applying a model to the preliminary image to assist in finding landmarks in the preliminary image; 
 extracting image based properties for the individual from the a finalized representation of the captured image of the ear; and 
 providing the extracted image based properties to a selection processor configured to select a customized HRTF dataset from a plurality of HRTF datasets that have been determined for a plurality of individuals. 
   
     
     
         12 . The processing device as recited in  claim 11 , wherein the image processing device is further configured for scaling the set of landmarks to provide a scaled representation of the preliminary image and further comprising a display screen for displaying the preliminary image to provide feedback to the user and wherein acquiring the captured image further includes generating a landmark template that is a visual guide overlaid in real-time onto the display screen of the image processing device to assist in alignment and which provides an indication of alignment to the user when the set of landmarks matched against the preliminary image are within an acceptable tolerance. 
     
     
         13 . The processing device as recited in  claim 12 , wherein the scaling is performed using from the captured image at least one of a conventionally sized reference object; a focal distance data associated with the focusing of the captured image; or a tragus length for the ear in the captured image. 
     
     
         14 . The processing device as recited in  claim 12 , wherein the customized HRTF dataset from the plurality of HRTF datasets is selected based on matching most closely the extracted image based properties to the image based properties associated with each of the HRTF datasets in the plurality of HRTF datasets. 
     
     
         15 . The processing device as recited in  claim 12 , wherein the model used in extracting image based properties for the captured image is an Active Shape Model and the Active Shape Model was previously trained on at least a plurality of images of individuals. 
     
     
         16 . The processing device as recited in  claim 12 , wherein multiple HRTF datasets are selected from the plurality of HRTF datasets by matching one or more of image based properties and the customized HRTF is generated by one of interpolation or perturbation relating to at least one of the multiple HRTF datasets, and wherein the image based properties comprise one or more of landmark distances, landmark arcs, landmark angles, landmark geometric relationships, a concha length, ear width, ear height, general physical dimensions of the ear, and a three dimensional representation of the ear. 
     
     
         17 . A system for generating customized HRTFs comprising:
 an image processing device configured for acquiring a captured image of at least one ear of an individual and for processing a preliminary image that is a preliminary version of the captured image to provide feedback to a user;   a device processor configured for:
 generating a set of landmarks by applying a model to the preliminary image to assist in finding landmarks in the preliminary image, wherein the landmarks correspond to control points; and 
 extracting image based properties for the individual from the captured image of the ear; 
   a selection processor for receiving the extracted image based properties and configured to select a customized HRTF dataset from a plurality of HRTF datasets that have been determined for a plurality of individuals wherein the device processor and the selection processor can be separate units or combined into one processor; and   a memory accessible by the selection processor and including the plurality of HRTF datasets, wherein the plurality of HRTF datasets is indexed by image based properties corresponding to an ear represented by each HRTF dataset in the plurality of HRTF datasets.   
     
     
         18 . The system as recited in  claim 17 , wherein acquiring a captured image further includes generating a landmark template that is a visual guide overlaid in real-time onto a display screen of the image processing device to assist in alignment and which provides an indication of alignment to the user when the landmarks matched against the preliminary image are within an acceptable tolerance 
     
     
         19 . The system as recited in  claim 18 , wherein the customized HRTF dataset from the plurality of HRTF datasets is selected based on matching most closely the extracted image based properties to the corresponding image based properties associated with each of the HRTF datasets in the plurality of HRTF datasets. 
     
     
         20 . The system as recited in  claim 17 , wherein multiple HRTF datasets are selected from the plurality of HRTF datasets by matching one or more of image based properties; and the customized HRTF is generated by one of interpolation or perturbation relating to at least one of the multiple HRTF datasets; and wherein the image based properties comprise one or more of landmark distances, landmark arcs, landmark angles, landmark geometric relationships, a concha length, ear width, ear height, general physical dimensions of the ear, and a three dimensional representation of the ear.