Ear biometric capture, authentication, and identification method and system

1. An ear-biometrics-based authentication and identification system comprising:

a smart phone;

an ear-image-capture subsystem within the smart phone that generates meta-data-associated images of a human ear by:

employing a front facing camera of the smart phone to record a series of camera frames as color images as the smart phone is pulled away from contact with a human ear,

employing one or more smart-phone sensors to generate additional data that is recorded as meta data that is associated with the color images, the one or more smart-phone sensors selected from among

a three-axis accelerometer that generates orientation angles from which a rotation of the phone from a vertical orientation is estimated,

a proximity sensor that generates indication of the distance from the human ear to the smart phone,

a light sensor that generates an indication of the ambient illumination in the environment, and

a touch-capacitive display screen that provides indications of points on the surface of the display screen at which the surface of the human ear contacts the display screen;

a template-generation subsystem that extracts features from the meta-data-associated images to generate a template that is stored in an electronic memory;

a template-comparison subsystem that

selects, by a modified-binary-search subsystem of the template-generation subsystem, color images from the recorded color images for feature extraction until an ear is detected in a color image, and

compares the generated template to previously generated and stored templates in order to identifier the user, the template-comparison subsystem generating a score that indicates how well the generated template matches a previously generated and stored template; and

a user interface through which a user can specify a threshold score that determines whether or not the generated template is determined by the template-comparison subsystem to correspond to the human ear from which the previously generated and stored template was generated.

2. The ear-biometrics-based authentication and identification system of claim 1 wherein color images sequentially adjacent to the color image in which an ear is identified are processed until no additional ears are detected.

3. The ear-biometrics-based authentication and identification system of claim 1 wherein the template-generation subsystem uses an OpenCV library cascade classifier, trained to quickly and efficiently detect vertically oriented ears.

4. The ear-biometrics-based authentication and identification system of claim 3 wherein the template-generation subsystem rotates a color image through a range of orientations to obtain a rotated color image in which a sub-image of an ear is vertically oriented.

5. The ear-biometrics-based authentication and identification system of claim 3 wherein the template-generation subsystem crops the color image in which an ear is detected to obtain a sub-image containing only the ear.

6. The ear-biometrics-based authentication and identification system of claim 3 wherein the template-generation subsystem applies a threshold to a color image remove hair and other large curved objects that can be mistaken for curves in an ear.

7. The ear-biometrics-based authentication and identification system of claim 3 wherein the template-generation subsystem denoises a color image to remove small curves that can be mistaken for curves of the ear.

8. The ear-biometrics-based authentication and identification system of claim 6 wherein the template-generation subsystem identifies local features of an ear in a color image by applying a bank of banana wavelets to the color image.

9. The ear-biometrics-based authentication and identification system of claim 1 wherein a template contains one or more feature vectors.