Apparatus and method for determining rotation angle of object, and computer-readable storage medium storing instructions to perform method for determining rotation angle of object

1 . A determining apparatus for determining a rotation angle of an object, comprising:

a memory; and

a processor executing one or more instructions stored in the memory, wherein the instructions, when executed by the processor, cause the processor to:

acquire an image,

convert the image into a black-and-white image,

extract a first feature point from the black-and-white image using a predetermined feature extraction algorithm,

match the first feature point of the object included in the image with a second feature point of a template by comparing the extracted first feature point with the second feature point within the template in which the second feature point for a predetermined object is previously stored,

determine the rotation angle of the object included in the image by comparing angle information of the first feature point with angle information of the second feature point, in which the rotation angle indicating an angle at which the object included in the image is rotated relative to an object included in the template,

recognize the object based on the template,

filter the angle information of the first feature point by identifying and removing data outlier values that deviate from a statistical distribution of the angle information, by using a statistical algorithm that calculates average and variance for the angle information of the first feature point to delete the data outlier that is a certain amount of variance away from the average, and

delete the data outlier from the image for correcting an error in the angle information of the first feature point, thereby improving accuracy of the determined rotation angle for real-time robotic manipulation of the object on a conveyor belt.

2 . The determining apparatus of claim 1 , wherein the object corresponds to the predetermined object, and

wherein the processor is configured to determine the rotation angle of the object by comparing angle information of a filtered feature point among first feature points with angle information of a second feature point matched with the filtered first feature point to determine a rotation angle of the object.

3 . The determining apparatus of claim 2 , wherein the processor is configured to determine the rotation angle of the object by calculating an angle difference between each filtered first feature point and the second feature point matched therewith, and calculating an average of calculated angle differences.

4 . The determining apparatus of claim 1 , wherein the processor is configured to extract angle information from the first feature point, and extract a descriptor on the basis of the angle information of the first feature point.

5 . The determining apparatus of claim 4 , wherein the processor is configured to specify the object included in the image by comparing the extracted descriptor with the descriptor previously stored in the template.

6 . The determining apparatus of claim 5 , wherein the processor is configured to specify that the recognized object corresponds to the predetermined object of the template, when a distance calculated through Hamming distance calculation for the extracted descriptor and the descriptor previously stored in the template is equal to or less than a preset threshold.

7 . A method for determining a rotation angle of an object performed by a determining apparatus, the method comprising:

acquiring an image;

converting the image into a black-and-white image;

extracting a first feature point from the black-and-white image using a predetermined feature extraction algorithm;

matching the first feature point of the object included in the image with a second feature point of a template by comparing the extracted first feature point with the second feature point within the template in which the second feature point for a predetermined object is previously stored;

determining the rotation angle of the object included in the image by comparing angle information of the first feature point with angle information of the second feature point,

recognizing the object based on the template,

filtering the angle information of the first feature point by identifying and removing data outlier values that deviate from a statistical distribution of the angle information, by using a statistical algorithm that calculates average and variance for the angle information of the first feature point to delete the data outlier that is a certain amount of variance away from the average, in which the rotation angle indicating an angle at which the object included in the image is rotated relative to an object included in the template, and

deleting the data outlier from the image for correcting an error in the angle information of the first feature point, thereby improving accuracy of the determined rotation angle for real-time robotic manipulation of the object on a conveyor belt.

8 . The determining method of claim 7 , wherein the object corresponds to the predetermined object, and

the determining includes comparing angle information of a filtered feature point among first feature points with angle information of a second feature point matched with the filtered first feature point to determine a rotation angle of the object.

9 . The determining method of claim 8 , wherein the determining includes calculating an angle difference between each filtered first feature point and the second feature point matched therewith, and averaging calculated angle differences, thus determining the rotation angle of the object included in the image.

10 . The determining method of claim 7 , wherein the extracting includes extracting angle information from the first feature point, and extracting a descriptor on the basis of the angle information of the first feature point.

11 . The determining method of claim 10 , wherein the matching includes specifying the object included in the image by comparing the extracted descriptor with the descriptor previously stored in the template.

12 . The determining method of claim 11 , wherein the matching includes specifying that the recognized object corresponds to the predetermined object of the template, when a distance calculated through Hamming distance calculation for the extracted descriptor and the descriptor previously stored in the template is equal to or less than a preset threshold.

13 . A non-transitory computer readable storage medium including computer executable instructions, wherein the instructions, when executed by a processor, cause the processor to perform a method for determining a rotation angle of an object, the method comprising:

acquiring an image;

converting the image into a black-and-white image;

extracting a first feature point from the black-and-white image using a predetermined feature extraction algorithm;

matching the first feature point of an object included in the image with a second feature point of a template by comparing the extracted first feature point with the second feature point within the template in which the second feature point for a predetermined object is previously stored; and

determining a rotation angle of the object included in the image by comparing angle information of the first feature point with angle information of the second feature point,

recognizing the object based on the template,

filtering the angle information of the first feature point by identifying and removing data outlier values that deviate from a statistical distribution of the angle information, by using a statistical algorithm that calculates average and variance for the angle information of the first feature point to delete the data outlier that is a certain amount of variance away from the average, in which the rotation angle indicating an angle at which the object included in the image is rotated relative to an object included in the template, and

deleting the data outlier from the image for correcting an error in the angle information of the first feature point, thereby improving accuracy of the determined rotation angle for real-time robotic manipulation of the object on a conveyor belt.

14 . The non-transitory computer readable storage medium of claim 13 , wherein the object corresponds to the predetermined object, and

the determining includes comparing angle information of a filtered feature point among first feature points with angle information of a second feature point matched with the filtered first feature point to determine a rotation angle of the object.