Image processing method and apparatus

1. An image processing method, comprising:

detecting a human face region in each image frame in a to-be-processed video containing a plurality of image frames;

locating a lip region in the human face region in each image frame;

generating a lip region map according to the located lip region in each image frame;

building a lip change map, including:

generating a first feature pixel column of the lip change map by selecting a column of pixels from the lip region map of a first image frame in the plurality of image frames;

generating a second feature pixel column of the lip change map by selecting a column of pixels from the lip region map of a second image frame in the plurality of image frames;

stitching the second feature pixel column laterally after the first feature pixel column in the lip change map;

generating one or more additional feature pixel columns of the lip change map by selecting a column of pixels from the lip region map of each of one or more additional image frames in the plurality of image frames; and

laterally stitching the one or more additional feature pixel columns after the second feature pixel column according to a time sequence of the one or more additional image frames; and

recognizing a lip movement according to a pattern feature of the lip change map.

2. The method according to claim 1 , wherein detecting a human face region in each image frame in the to-be-processed video, and locating a lip region in the human face region in each image frame comprises:

parsing the to-be-processed video to obtain the plurality of image frames;

detecting a human face region in each image frame by using a human face recognition algorithm; and

locating the lip region in the human face region of each image frame by using a human face alignment algorithm.

3. The method according to claim 1 , wherein selecting a column of pixels from a lip region map comprises:

selecting a position in the lip region map;

drawing a longitudinal axis passing through the selected position; and

selecting pixels along the longitudinal axis in the lip region map.

4. The method according to claim 3 , wherein the selected position is a position of a central pixel in the lip region map.

5. The method according to claim 1 , wherein recognizing a lip movement according to a pattern feature of the lip change map comprises:

calculating the pattern feature of the lip change map, the pattern feature comprising at least one of an LBP (Local Binary Pattern) feature or or a HOG (Histogram of Oriented Gradient) feature; and

classifying the pattern feature by using a classification algorithm, to obtain a lip movement recognition result, the lip movement recognition result being a determination on a lip movement occurrence or no lip movement occurrence.

6. A lip movement recognition system, comprising:

a memory, storing a group of program code; and

a processor, configured to execute the program code to perform:

detecting a human face region in each image frame in a to-be-processed video containing a plurality of image frames;

locating a lip region in the human face region in each image frame;

generating a lip region map according to the located lip region in each image frame;

building a lip change map, including:

generating a first feature pixel column of the lip change map by selecting a column of pixels from the lip region map of a first image frame in the plurality of image frames;

generating a second feature pixel column of the lip change map by selecting a column of pixels from the lip region map of a second image frame in the plurality of image frames;

stitching the second feature pixel column laterally after the first feature pixel column in the lip change map;

generating one or more additional feature pixel columns of the lip change map by selecting a column of pixels from the lip region map of each of one or more additional image frames in the plurality of image frames; and

laterally stitching the one or more additional feature pixel columns after the second feature pixel column according to a time sequence of the one or more additional image frames; and

recognizing a lip movement according to a pattern feature of the lip change map.

7. The lip movement recognition system according to claim 6 , wherein detecting a human face region in each image frame in the to-be-processed video, and locating a lip region in the human face region comprises:

parsing the to-be-processed video to obtain the plurality of image frames;

detecting a human face region in each image frame by using a human face recognition algorithm; and

locating the lip region in the human face region of each image frame by using a human face alignment algorithm.

8. The lip movement recognition system according to claim 6 , wherein selecting a column of pixels from a lip region map comprises:

selecting a position in the lip region map;

drawing a longitudinal axis passing through the selected position; and

selecting column pixels along the longitudinal axis in the lip region map.

9. The lip movement recognition system according to claim 8 , the position being a position of a central pixel in the lip region map.

10. The lip movement recognition system according to claim 6 , wherein recognizing a lip movement according to a pattern feature of the lip change map comprises:

calculating the pattern feature of the lip change map, the pattern feature comprising at least one of an LBP feature or a HOG feature; and

classifying the pattern feature by using a preset classification algorithm, to obtain a lip movement recognition result, the lip movement recognition result being a determination on a lip movement occurs or no lip movement occurs.

11. A non-transitory computer-readable storage medium storing program instructions which, when being executed by at least one processor, cause the at least one processor to perform:

detecting a human face region in each image frame in a to-be-processed video containing a plurality of image frames;

locating a lip region in the human face region in each image frame;

generating a lip region map according to the located lip region in each image frame;

building a lip change map, including:

generating a first feature pixel column of the lip change map by selecting a column of pixels from the lip region map of a first image frame in the plurality of image frames;

generating a second feature pixel column of the lip change map by selecting a column of pixels from the lip region map of a second image frame in the plurality of image frames;

stitching the second feature pixel column laterally after the first feature pixel column in the lip change map;

generating one or more additional feature pixel columns of the lip change map by selecting a column of pixels from the lip region map of each of one or more additional image frames in the plurality of image frames; and

laterally stitching the one or more additional feature pixel columns after the second feature pixel column according to a time sequence of the one or more additional image frames; and

recognizing a lip movement according to a pattern feature of the lip change map.

12. The computer-readable medium according to claim 11 , wherein detecting a human face region in each image frame in the to-be-processed video, and locating a lip region in the human face region in each image frame comprises:

parsing the to-be-processed video to obtain the plurality of image frames;

detecting a human face region in each image frame by using a human face recognition algorithm; and

locating the lip region in the human face region of each image frame by using a human face alignment algorithm.

13. The computer-readable medium according to claim 11 , wherein selecting a column of pixels from a lip region map comprises:

selecting a position in the lip region map;

drawing a longitudinal axis passing through the selected position; and

selecting pixels along the longitudinal axis in the lip region map.

14. The computer-readable medium according to claim 13 , wherein the selected position is a position of a central pixel in the lip region map.

15. The computer-readable medium according to claim 11 , wherein recognizing a lip movement according to a pattern feature of the lip change map comprises:

calculating the pattern feature of the lip change map, the pattern feature comprising at least one of an LBP (Local Binary Pattern) feature or a HOG (Histogram of Oriented Gradient) feature; and

classifying the pattern feature by using a classification algorithm, to obtain a lip movement recognition result, the lip movement recognition result being a determination on a lip movement occurrence or no lip movement occurrence.

16. The method according to claim 1 , wherein the lip change map is a two-dimensional map.

17. The method according to claim 5 , wherein the classification algorithm is one of: a Bayesian algorithm, a logistic regression algorithm, and an SVM (Support Vector Machine) algorithm.