Method and apparatus for processing image

A non-transitory computer-readable medium having a computer program stored thereon,

• wherein the computer program, when executed on a computer, causes the computer to implement:
• receiving a scanned image;
• and
• processing the received image through an octave convolution-based neural network to output a high-quality image and an edge image for the received image,
• wherein the octave convolution-based neural network comprises a plurality of octave encoder blocks and a plurality of octave decoder blocks,
• wherein each octave encoder block comprises an octave convolutional layer, and is configured to output a high-frequency feature map and a low-frequency feature map for an image inputted to each octave encoder block, and
• wherein a first octave encoder block, which is a first block among the plurality of octave encoder blocks, is configured to:
• transfer the low-frequency feature map outputted from the first octave encoder block to a second octave encoder block, which is a second block among the plurality of octave encoder blocks, without transferring the high-frequency feature map outputted from the first octave encoder block to the second octave encoder block, and
• transfer the high-frequency feature map outputted from the first octave encoder block to a first octave decoder block, which is the last block among the plurality of octave decoder blocks, without transferring the low-frequency feature map outputted from the first octave encoder block to the first octave decoder block.

An image processing device, comprising:

• a memory:
• and
• at least one processor configured to be connected to the memory and execute computer readable instructions stored in the memory,
• wherein the at least one processor is configured to:
• receive a scanned image;
• and
• process the received image through an octave convolution-based neural network to
• output a high-quality image and an edge image for the received image,
• wherein the octave convolution-based neural network comprises a plurality of octave encoder blocks and a plurality of octave decoder blocks,
• wherein each octave encoder block comprises an octave convolutional layer, and is configured to output a high-frequency feature map and a low-frequency feature map for an image inputted to each octave encoder block, and
• wherein a first octave encoder block, which is a first block among the plurality of octave encoder blocks, is configured to:
• transfer the low-frequency feature map outputted from the first octave encoder block to a second octave encoder block, which is a second block among the plurality of octave encoder blocks, without transferring the high-frequency feature map outputted from the first octave encoder block to the second octave encoder block, and
• transfer the high-frequency feature map outputted from the first octave encoder block to a first octave decoder block, which is the last block among the plurality of octave decoder blocks, without transferring the low-frequency feature map outputted from the first octave encoder block to the first octave decoder block.