Image processing method and device, classifier training method, and readable storage medium

1. An image processing method, comprising:

extracting a characteristic vector in an image to be recognized;

based on the characteristic vector of the image to be recognized, acquiring a predicted score value of the image to be recognized; and

based on the predicted score value, determining a category of an image information of the image to be recognized;

wherein the image to be recognized is a face image, and the image information is a facial expression;

wherein the step of, based on the characteristic vector of the image to be recognized, acquiring the predicted score value of the image to be recognized comprises:

by using a random-forest classifier, based on the characteristic vector of the image to be recognized, acquiring the predicted score value of the image to be recognized; and

the method further comprises:

based on characteristic vectors and grading values of image data in a training set comprised in a dataset, training the random-forest classifier;

wherein the dataset comprises the plurality of image data, and each of the image data is marked with a grading value;

the grading value of each of the image data is determined according to a category of image information comprised in the each of the image data, and different categories of the image information correspond to different grading values;

the image data are face-image data; and

the training set comprises a fifth quantity of the image data in the dataset.

2. The image processing method according to claim 1 , wherein the step of extracting the characteristic vector in the image to be recognized comprises:

by using a Gabor filter, acquiring an image-feature response diagram of the image to be recognized; and

extracting the characteristic vector of the image to be recognized from the image-feature response diagram;

wherein the Gabor filter comprises a first quantity of dimensions and a second quantity of directions;

the image-feature response diagram comprises features of the image information of the image to be recognized; and

the first quantity of dimensions are less than 4 dimensions.

3. The image processing method according to claim 2 , wherein the method further comprises, according to an accuracy rate of recognition of the image information by the Gabor filter with a third quantity of dimensions and a fourth quantity of directions, selecting the first quantity of dimensions and the second quantity of directions of the Gabor filter;

wherein the first quantity of dimensions are 2 dimensions, and the second quantity of directions are 3 directions.

4. The image processing method according to claim 1 , wherein the step of, based on the characteristic vectors and the grading values of the image data in the training set, training the random-forest classifier comprises:

by using a K-fold cross-validation mode, training the random-forest classifier;

wherein K represents K parts that the image data in the training set are divided into, and K is an integer greater than 0.

5. The image processing method according to claim 1 , further comprising:

based on characteristic vectors and grading values of image data in a test set comprised in the dataset, checking the random-forest classifier that has been trained;

wherein the test set comprises a sixth quantity of the image data; and

the fifth quantity is greater than the sixth quantity.

6. The image processing method according to claim 1 , wherein the categories of the image information include disgust, anger, calmness, delight and astonishment;

wherein the disgust corresponds to a grading value of 10-19, the anger corresponds to a grading value of 20-29, the calmness corresponds to a grading value of 30-39, the delight corresponds to a grading value of 40-49, and the astonishment corresponds to a grading value of 50-59.

7. The image processing method according to claim 6 , wherein the step of, based on the predicted score value, determining the category of the image information of the image to be recognized comprises:

determining a threshold interval that the predicted score value of the image to be recognized falls into; and

according to the threshold interval that the predicted score value of the image to be recognized falls into, determining the category of the image information of the image to be recognized.

8. An image processing device, comprising:

a processor; and

a memory, wherein the memory stores one or more computer instructions, and the one or more computer instructions, when executed by the processor, causes the processor to perform operations of:

extracting a characteristic vector in an image to be recognized;

based on the characteristic vector of the image to be recognized, acquiring a predicted score value of the image to be recognized; and

based on the predicted score value, determining a category of an image information of the image to be recognized;

wherein the image to be recognized is a face image, and the image information is a facial expression;

wherein the operation of, based on the characteristic vector of the image to be recognized, acquiring the predicted score value of the image to be recognized comprises:

by using a random-forest classifier, based on the characteristic vector of the image to be recognized, acquiring the predicted score value of the image to be recognized; and

the operations further comprise:

based on characteristic vectors and grading values of image data in a training set comprised in a dataset, training the random-forest classifier;

wherein the dataset comprises the plurality of image data, and each of the image data is marked with a grading value;

the grading value of each of the image data is determined according to a category of image information comprised in the each of the image data, and different categories of the image information correspond to different grading values;

the image data are face-image data; and

the training set comprises a fifth quantity of the image data in the dataset.

9. A training method of a random-forest classifier, comprising:

based on characteristic vectors and grading values of image data in a training set comprised in a dataset, training the random-forest classifier;

wherein the dataset comprises the plurality of image data, and each of the plurality of image data is marked with a grading value;

the grading value of each of the image data is determined according to a category of image information comprised in the each of the image data, and different categories of the image information correspond to different grading values;

the image data are face-image data, and the image information is a facial expression; and

the training set comprises a fifth quantity of the image data in the dataset.

10. The training method according to claim 9 , wherein the step of, based on the characteristic vectors and the grading values of the image data in the training set, training the random-forest classifier comprises:

by using a K-fold cross-validation mode, training the random-forest classifier;

wherein K represents K parts that the image data in the training set are divided into, and K is an integer greater than 0.

11. The training method according to claim 10 , further comprising:

based on characteristic vectors and grading values of image data in a test set comprised in the dataset, checking the random-forest classifier that has been trained;

wherein the test set comprises a sixth quantity of image data in the dataset; and

the fifth quantity is greater than the sixth quantity.

12. The training method according to claim 9 , wherein the categories of the image information include disgust, anger, calmness, delight and astonishment;

wherein the disgust corresponds to a grading value of 10-19, the anger corresponds to a grading value of 20-29, the calmness corresponds to a grading value of 30-39, the delight corresponds to a grading value of 40-49, and the astonishment corresponds to a grading value of 50-59.

13. A nonvolatile computer-readable storage medium, wherein the computer-readable storage medium stores a computer-readable instruction, and the computer-readable instruction, when executed by a computer, causes the computer to implement the image processing method according to claim 1 .

14. The image processing device according to claim 8 , wherein the one or more computer instructions, when executed by the processor, further causes the processor to perform operations of:

by using a Gabor filter, acquiring an image-feature response diagram of the image to be recognized; and

extracting the characteristic vector of the image to be recognized from the image-feature response diagram;

wherein the Gabor filter comprises a first quantity of dimensions and a second quantity of directions;

the image-feature response diagram comprises features of the image information of the image to be recognized; and

the first quantity of dimensions are less than 4 dimensions.

15. The image processing device according to claim 14 , wherein the one or more computer instructions, when executed by the processor, further causes the processor to perform an operation of:

according to an accuracy rate of recognition of the image information by the Gabor filter with a third quantity of dimensions and a fourth quantity of directions, selecting the first quantity of dimensions and the second quantity of directions of the Gabor filter;

wherein the first quantity of dimensions are 2 dimensions, and the second quantity of directions are 3 directions.

16. The nonvolatile computer-readable storage medium according to claim 13 , wherein the step of extracting the characteristic vector in the image to be recognized comprises:

by using a Gabor filter, acquiring an image-feature response diagram of the image to be recognized; and

extracting the characteristic vector of the image to be recognized from the image-feature response diagram;

wherein the Gabor filter comprises a first quantity of dimensions and a second quantity of directions;

the image-feature response diagram comprises features of the image information of the image to be recognized; and

the first quantity of dimensions are less than 4 dimensions.