| CPC G06V 10/273 (2022.01) [G06V 30/155 (2022.01); G06V 30/10 (2022.01)] | 6 Claims |
|
1. An apparatus comprising:
at least one memory that stores instructions; and
at least one processor that executes the instructions to perform:
generating a first binary image from an input image by changing first pixels of the input image into first black pixels and changing second pixels of the input image into first white pixels, wherein luminances of the first pixels are lower than a first threshold, and wherein the second pixels are other than the first pixels in the input image;
identifying character images from the first black pixels in the first binary image;
generating a second binary image from the input image by changing third pixels of the input image into second black pixels and changing fourth pixels of the input image into second white pixels, wherein luminances of the third pixels are higher than the first threshold and are lower than luminance of a background in the input image, wherein the second binary image is generated from the input image without using the generated first binary image;
determining whether each of black pixel clusters detected based on the second black pixels that connect with each other in the second binary image is a target, wherein the detected black pixel clusters include at least one of a first black pixel cluster, a second black pixel cluster and a third black pixel cluster,
wherein it is determined that the first black pixel cluster in the second binary image is the target in a case where a position of an area of the first black pixel cluster in the second binary image does not overlap a position of any area of the identified character images in the first binary image and where the position of the area of the first black pixel cluster in the second binary image is not between positions of any two areas of the identified character images in the first binary image,
wherein it is determined that the second black pixel cluster in the second binary image is not the target in a case where a position of an area of the second black pixel cluster in the second binary image overlaps a position of any area of the identified character images in the first binary image, and
wherein it is determined that the third black pixel cluster in the second binary image is not the target in a case where the position of the area of the third black pixel cluster in the second binary image is between positions of any two areas of the identified character images in the first binary image; and
performing character recognition processing on the identified character images and the first black pixel cluster that has been determined to be the target.
|