Method for acquiring object information and apparatus for performing same

1. A method for acquiring an object information, comprising:

obtaining an input image acquired by capturing a sea;

obtaining a noise level of the input image, the noise level reflecting an environmental noise of the input image;

when the noise level indicates a noise lower than a predetermined level, acquiring an object information related to an obstacle included in the input image from the input image by using a first artificial neural network wherein the first artificial neural network outputs, from an image, an object information related to at least one selected from the group of a distance and a type of an object included in the image; and

when the noise level indicates a noise higher than the predetermined level, obtaining a noise-reduced image of which the environmental noise is reduced from the input image by using a second artificial neural network trained to generate, from a first image including a noise, a second image of which the noise included in the first image is reduced, and acquiring an object information related to an obstacle included in the sea from the noise-reduced image by using the first artificial neural network, wherein the second artificial neural network is trained by using a first error calculated from a third artificial neural network discriminating the second image and a reference image corresponding to the first image and having a noise lower than the first image.

2. The method of claim 1 , wherein the first artificial neural network is trained based on an output data and a labeling data corresponding to the output data, the output data including a first classification value determined by reflecting the type and corresponding to a sea, and a second classification value determined by reflecting the distance and the type and corresponding to an obstacle.

3. The method of claim 1 , wherein the environmental noise includes at least one selected from the group of a rain, a fog and a cloud.

4. The method of claim 1 , wherein the obtaining the input image comprises selecting a portion of a plurality of images.

5. The method of claim 4 , wherein the selecting is generating an image by compositing an image having the highest illuminance among the plurality of images and an image having the lowest illuminance among the plurality of images.

6. The method of claim 1 , wherein the obtaining the input image comprises RGB normalizing the input image.

7. The method of claim 1 , wherein the obtaining the noise level comprises measuring an average focus measure of the input image.

8. The method of claim 1 , wherein the second artificial neural network is trained to minimize the first error.

9. The method of claim 1 , wherein the second artificial neural network is trained to minimize a second error calculated based on a comparison of a third image outputted from a fourth artificial neural network and the first image, the fourth artificial neural network generating the third image having a noise higher than the second image inputted to the fourth artificial neural network.

10. A method for acquiring an object information, comprising:

obtaining an input image acquired by capturing a sea;

obtaining a noise level of the input image, the noise level reflecting an environmental noise of the input image;

when the noise level indicates a noise lower than a predetermined level, acquiring an object information related to an obstacle included in the input image from the input image by using a first artificial neural network wherein the first artificial neural network outputs, from an image, an object information related to at least one selected from the group of a distance and a type of an object included in the image; and

when the noise level indicates a noise higher than the predetermined level, generating a noise-reducing mask from the input image by using a second artificial neural network trained to generate an image mask from a first image including a noise, obtaining a generated image of which the environmental noise is reduced from the input image by using the noise-reducing mask, and acquiring an object information related to an obstacle included in the sea from the generated image by using the first artificial neural network, wherein the second artificial neural network is trained by using a first error calculated from a third artificial neural network discriminating a second image and a reference image corresponding to the first image and having a noise lower than the first image, the second image being generated based on an image mask generated from the second artificial neural network and the first image.

11. The method of claim 10 , wherein the first artificial neural network is trained based on an output data and a labeling data corresponding to the output data, the output data including a first classification value determined by reflecting the type and corresponding to a sea, and a second classification value determined by reflecting the distance and the type and corresponding to an obstacle.

12. The method of claim 10 , wherein:

the input image includes a first input image and a second input image that is a subsequent frame of the first input image, and

when the noise level indicates a noise higher than the predetermined level and a difference between a first characteristic of the first input image and a second characteristic of the second input image is smaller than a predetermined reference value, generating a first noise-reducing mask from the first input image, obtaining a first generated image of which the environmental noise is reduced from the first input image by using the first image mask, and acquiring an object information related to an obstacle included in the first input image from the first generated image by using the first artificial neural network, and

when the noise level indicates a noise higher than the predetermined level and the difference between the first characteristic and the second characteristic is higher than the predetermined reference value, generating a second noise-reducing mask from the second input image, obtaining a second generated image of which the environmental noise is reduced from the second input image by using the second image mask, and acquiring an object information related to an obstacle included in the sea from the second generated image by using the first artificial neural network.

13. The method of claim 12 , wherein each of the first characteristic and the second characteristic includes a noise level and a RGB value of a pixel.

14. The method of claim 10 , wherein the second artificial neural network is trained to minimize a second error calculated based on a comparison of a third image outputted from a fourth artificial neural network and the first image, the fourth artificial neural network generating the third image having a noise higher than the second image inputted to the fourth artificial neural network.

15. The method of claim 10 , wherein the noise-reducing mask is maintained for a predetermined frame.

16. A device for acquiring an object information, comprising:

a camera mounted on a vessel and capturing a sea; and

a controller configured to:

obtain an input image captured by the camera;

obtain a noise level of the input image, the noise level reflecting an environmental noise of the input image;

when the noise level indicates a noise lower than a predetermined level, acquiring an object information related to an obstacle included in the input image from the input image by using a first artificial neural network wherein the first artificial neural network outputs, from an image, an object information related to at least one selected from the group of a distance and a type of an object included in the image; and

when the noise level indicates a noise higher than the predetermined level, generating a noise-reducing mask from the input image by using a second artificial neural network trained to generate an image mask from a first image including a noise, obtaining a generated image of which the environmental noise is reduced from the input image by using the noise-reducing mask, and acquiring an object information related to an obstacle included in the sea from the generated image by using the first artificial neural network, wherein the second artificial neural network is trained by using a first error calculated from a third artificial neural network discriminating a second image and a reference image corresponding to the first image and having a noise lower than the first image, the second image being generated based on an image mask generated from the second artificial neural network and the first image.