User authentication method and apparatus

1. A user authentication method comprising:

extracting, by a processor, a facial landmark from each of a first input image and a second input image;

generating, by the processor, a first sparse code of the facial landmark extracted from the first input image and a second sparse code of the facial landmark extracted from the second input image; and

determining, by the processor, whether a user is to be authenticated based on the first sparse code of the first input image and the second sparse code of the second input image, the determining including,

calculating a similarity between the first sparse code and the second sparse code,

determining a final similarity based on the calculated similarity, and

determining whether a user represented in the first input image is similar to a user represented in the second input image based on whether the final similarity satisfies a desired condition.

2. The method of claim 1 , wherein the generating comprises:

determining a first feature vector set associated with the facial landmark extracted from the first input image and a second feature vector set associated with the facial landmark extracted from the second input image, the first feature vector set and the second feature vector set being generated based on learning data.

3. The method of claim 2 , wherein the generating further comprises:

generating the first sparse code using the first feature vector set; and

generating the second sparse code using the second feature vector set.

4. The method of claim 2 , wherein the generating further comprises:

generating a first feature vector for the facial landmark extracted from the first input image based on an area surrounding the facial landmark;

generating a second feature vector for the facial landmark extracted from the second input image based on an area surrounding the facial landmark;

generating the first sparse code for the facial landmark extracted from the first input image using the first feature vector set and the generated first feature vector; and

generating the second sparse code for the facial landmark extracted from the second input image using the second feature vector set and the generated second feature vector.

5. The method of claim 1 , wherein the extracting extracts a plurality of facial landmarks from the first input image and the second input image, and the generating comprises:

generating the first sparse code for each of the plurality of facial landmarks extracted from the first input image; and

generating the second sparse code for each of the plurality of facial landmarks extracted from the second input image.

6. The method of claim 1 , wherein the calculating comprises:

calculating the similarity between the first sparse code and the second sparse code through one of a cosine distance measurement and a chi-square distance measurement.

7. The method of claim 1 , wherein the facial landmark comprises at least one of both ends of an eye, both ends of an eyebrow, a tip of a nose, and both corners of a mouth.

8. A non-transitory computer-readable medium storing program code that, which when executed by a processor, configures the processor to performs the method of claim 1 .

9. A user authentication method comprising:

extracting, by a processor, a facial landmark from a first input image;

generating, by the processor, a first sparse code for the facial landmark using a first feature vector set generated based on learning data;

generating, by the processor, a transformed sparse code for the facial landmark using a second feature vector set generated based on a second input image and the learning data, the second feature vector set including a feature vector included in the first feature vector set, the feature vector generated based on a facial landmark extracted from the second input image; and

determining, by the processor, whether a user is to be authenticated based on the first sparse code generated using the first feature vector set and the transformed sparse code.

10. The method of claim 9 , wherein the generating the first sparse code comprises:

generating a first feature vector for the extracted facial landmark based on an area surrounding the facial landmark; and

generating the first sparse code for the facial landmark using the first feature vector set and the generated first feature vector.

11. The method of claim 9 , wherein the extracting extracts a plurality of facial landmarks, and the generating the first sparse code comprises:

generating the first sparse code for each of the plurality of extracted facial landmarks using the first feature vector set for each of the plurality of facial landmarks.

12. The method of claim 9 , wherein the determining comprises:

calculating a similarity between the first sparse code generated and the transformed sparse code;

determining a final similarity based on the calculated similarity; and

determining whether a user represented in the first input image is similar to a user represented in the second input image based on whether the final similarity satisfies a desired condition.

13. A user authentication apparatus comprising:

at least one processor configured to,

extract a facial landmark from each of a first input image and a second input image,

generate a first sparse code corresponding to the facial landmark extracted from the first input image,

generate a second sparse code corresponding to the facial landmark extracted from the second input image, and

determine whether a user is to be authenticated based on the first sparse code and the second sparse code by,

calculating a similarity between the first sparse code and the second sparse code,

determining a final similarity based on the calculated similarity, and

determining whether a user represented in the first input image is similar to a user represented in the second input image, based on whether the final similarity satisfies a desired condition.

14. The apparatus of claim 13 , wherein:

the at least one processor is configured to generate the first sparse code for the facial landmark extracted from the first input image using a first feature vector set, the first feature vector set generated based on learning data; and

the at least one processor is configured to generate the second sparse code for the facial landmark extracted from the second input image using a second feature vector set, the second feature vector set generated based on the learning data.

15. The apparatus of claim 14 , wherein:

the at least one processor is further configured to generate a first feature vector for the facial landmark based on an area surrounding the facial landmark;

the at least one processor is configured to generate the first sparse code for the facial landmark using the first feature vector set and the generated first feature vector;

the at least one processor is further configured to generate a second feature vector for the facial landmark based on an area surrounding the facial landmark; and

the at least one processor is configured to generate the second sparse code for the facial landmark using the second feature vector set and the generated first feature vector.