Image recognition method for performing image recognition utilizing convolution filters

1. An image recognition method performed by a computer of an image recognition device, comprising:

receiving an image;

acquiring first processing result information including values of a first processing result of first convolution processing at positions of a plurality of pixels that constitute the image by performing the first convolution processing on the image by using a first convolution filter;

acquiring second processing result information including values of a second processing result of second convolution processing at positions of a plurality of pixels that constitute the image by performing the second convolution processing on the image by using a second convolution filter that is different from the first convolution filter in resolution, scale parameter or color to be processed;

determining one feature for each of the positions of the plurality of pixels on the basis of the values of the first processing result and the second processing result at the positions of the plurality of pixels included in the first processing result information and the second processing result information and outputting the determined feature for each of the positions of the plurality of pixels;

performing recognition processing on the basis of the determined feature for each of the positions of the plurality of pixels; and

outputting recognition processing result information obtained by performing the recognition processing.

2. The image recognition method according to claim 1 , wherein

a resolution of the first convolution filter is higher than a resolution of the second convolution filter.

3. The image recognition method according to claim 1 , wherein

in the outputting the determined feature for each of the positions of the plurality of pixels, one feature for the position is determined by selecting, as the feature at the position, a maximum value among the values of the first processing result and the second processing result at the position, and outputting the determined feature for the position.

4. The image recognition method according to claim 1 , wherein

in the outputting the determined feature for each of the positions of the plurality of pixels, one feature for the position is determined by calculating a median or an average of the values of the first processing result and the second processing result at the position and determining the calculated value as the feature at the position, and outputting the determined feature at the position.

5. The image recognition method according to claim 1 , wherein

in the outputting the determined feature for each of the positions of the plurality of pixels, subsampling processing is performed for determining, for each region including a plurality of adjacent pixels, any of the features at positions of the plurality of pixels included in the region as a representative feature representing the region; and the recognition processing is performed on the basis of the representative feature determined in the subsampling processing.

6. The image recognition method according to claim 5 , wherein

in the subsampling processing, a feature whose value is largest among the features at the positions of the plurality of pixels included in the region is determined as the representative feature.

7. The image recognition method according to claim 1 , wherein

at least one of the receiving the image, the acquiring the first processing result information, the acquiring the second processing result information, the determining the one feature for each of the positions of the plurality of pixels and the outputting the determined feature for each of the positions of the plurality of pixels, the performing the recognition processing and the outputting the recognition processing result information is performed by a processor provided in the computer of the image recognition device.

8. An image recognition device comprising:

a processor; and

a memory that stores a program,

wherein, the program causes the processor to execute:

receiving an image;

acquiring first processing result information including values of a first processing result of first convolution processing at positions of a plurality of pixels that constitute the image by performing the first convolution processing on the image by using a first convolution filter;

acquiring second processing result information including values of a second processing result of second convolution processing at positions of a plurality of pixels that constitute the image by performing the second convolution processing on the image by using a second convolution filter that is different from the first convolution filter in resolution, scale parameter or color to be processed;

determining one feature for each of the positions of the plurality of pixels on the basis of the values of the first processing result and the second processing result at the positions of the plurality of pixels included in the first processing result information and the second processing result information and outputting the determined feature for each of the positions of the plurality of pixels;

performing recognition processing on the basis of the determined feature for each of the positions of the plurality of pixels that is output; and

outputting recognition processing result information obtained by performing the recognition processing.

9. The image recognition method according to claim 8 , wherein

a resolution of the first convolution filter is higher than a resolution of the second convolution filter.

10. The image recognition method according to claim 8 , wherein

in the outputting the determined feature for each of the positions of the plurality of pixels, one feature for the position is determined by selecting, as the feature at the position, a maximum value among the values of the first processing result and the second processing result at the position, and outputting the determined feature for the position.

11. The image recognition method according to claim 8 , wherein

in the outputting the determined feature for each of the positions of the plurality of pixels, one feature for the position is determined by calculating a median or an average of the values of the first processing result and the second processing result at the position and determining the calculated value as the feature at the position, and outputting the determined feature at the position.

12. The image recognition method according to claim 8 , wherein

in the outputting the determined feature for each of the positions of the plurality of pixels, subsampling processing is performed for determining, for each region including a plurality of adjacent pixels, any of the features at positions of the plurality of pixels included in the region as a representative feature representing the region; and the recognition processing is performed on the basis of the representative feature determined in the subsampling processing.

13. The image recognition method according to claim 12 , wherein

in the subsampling processing, a feature whose value is largest among the features at the positions of the plurality of pixels included in the region is determined as the representative feature.

14. The image recognition method according to claim 8 , wherein

at least one of the receiving the image, the acquiring the first processing result information, the acquiring the second processing result information, the determining the one feature for each of the positions of the plurality of pixels and the outputting the determined feature for each of the positions of the plurality of pixels, the performing the recognition processing and outputting the recognition processing result information is performed by a processor provided in the computer of the image recognition device.

15. A non-transitory computer-readable recording medium storing a program for causing a computer of an image recognition device to:

receive an image;

acquire first processing result information including values of a first processing result of first convolution processing at positions of a plurality of pixels that constitute the image by performing the first convolution processing on the image by using a first convolution filter;

acquire second processing result information including values of a second processing result of second convolution processing at positions of a plurality of pixels that constitute the image by performing the second convolution processing on the image by using a second convolution filter that is different from the first convolution filter in resolution, scale parameter or color to be processed;

determine one feature for each of the positions of the plurality of pixels on the basis of the values of the first processing result and the second processing result at the positions of the plurality of pixels included in the first processing result information and the second processing result information and output the determined feature for each of the positions of the plurality of pixels;

perform recognition processing on the basis of the determined feature for each of the positions of the plurality of pixels; and

output recognition processing result information obtained by performing the recognition processing.

16. The image recognition method according to claim 15 , wherein

a resolution of the first convolution filter is higher than a resolution of the second convolution filter.

17. The image recognition method according to claim 15 , wherein

in the output of the determined feature for each of the positions of the plurality of pixels, one feature for the position is determined by selecting, as the feature at the position, a maximum value among the values of the first processing result and the second processing result at the position, and outputting the determined feature for the position.

18. The image recognition method according to claim 15 , wherein

in the output of the determined feature for each of the positions of the plurality of pixels, one feature for the position is determined by calculating a median or an average of the values of the first processing result and the second processing result at the position and determining the calculated value as the feature at the position, and outputting the determined feature at the position.

19. The image recognition method according to claim 15 , wherein

in the output of the determined feature for each of the positions of the plurality of pixels, subsampling processing is performed for determining, for each region including a plurality of adjacent pixels, any of the features at positions of the plurality of pixels included in the region as a representative feature representing the region; and the recognition processing is performed on the basis of the representative feature determined in the subsampling processing.

20. The image recognition method according to claim 19 , wherein

in the subsampling processing, a feature whose value is largest among the features at the positions of the plurality of pixels included in the region is determined as the representative feature.