Estimating gaze direction

1. An information processing apparatus comprising:

a detection unit adapted to detect a face region and an eye region of an object from an obtained image of the object;

a generation unit adapted to generate normalized images of a high-resolution image and low-resolution image from the obtained image of the face region based on a positional relationship between the face region and the eye region in the image;

a first extraction unit adapted to extract a first feature amount indicating a direction of a face existing in the face region from the low-resolution image;

a feature point detection unit adapted to detect a position of a feature point indicating the eye region in the normalized image of the high-resolution image;

a clipping unit adapted to clip the eye region in the normalized image of the high-resolution image based on the position of the feature point detected by said feature point detection unit;

an enlargement/reduction unit adapted to enlarge or reduce the eye region clipped by said clipping unit to make the eye region have a predetermined size;

a second extraction unit adapted to extract a second feature amount indicating a direction of an eye existing in the eye region from the eye region having the predetermined size which is enlarged or reduced by said enlargement/reduction unit; and

an estimation unit adapted to estimate a gaze direction of the object from the first feature amount and the second feature amount.

2. The apparatus according to claim 1 , wherein said clipping unit is configured to clip, as the eye region, a rectangular region including a feature point indicating a position of at least one of an external canthus, a position of an internal canthus, a position of an upper eyelid, and a position of lower eyelid.

3. The apparatus according to claim 1 , wherein said second extraction unit is configured to extract, as the second feature amount, one of a first derivative value and a second derivative value of pixel values constituting the eye region enlarged or reduced by said enlargement/reduction unit and having the predetermined size.

4. The apparatus according to claim 1 , wherein said first extraction unit is configured to extract, as the first feature amount, one of a first derivative value and a second derivative value of pixel values constituting the low-resolution image.

5. An information processing method comprising:

a detection step of detecting a face region and an eye region of an object from an obtained image of the object;

a generation step of generating normalized images of a high-resolution image and low-resolution image from the obtained image of the face region based on a positional relationship between the face region and the eye region in the image;

a first extraction step of extracting a first feature amount indicating a direction of a face existing in the face region from the low-resolution image;

a feature point detection step of detecting a position of a feature point indicating the eye region in the normalized image of the high-resolution image;

a clipping step of clipping the eye region in the normalized image of the high-resolution image based on the position of the feature point detected in said feature point detection step;

an enlargement/reduction step of enlarging or reducing the eye region clipped in said clipping step to make the eye region have a predetermined size;

a second extraction step of extracting a second feature amount indicating a direction of an eye existing in the eye region from the eye region having the predetermined size which is enlarged or reduced in said enlargement/reduction step; and

an estimation step of estimating a gaze direction of the object from the first feature amount and the second feature amount.

6. A non-transitory computer-readable storage medium storing a computer program for causing a computer to execute an information processing method defined in claim 5 .