20230169628. APPARATUS AND METHOD FOR IMAGE OUTPAINTING BASED ON LEARNING OF A SINGLE INPUT IMAGE simplified abstract (ELECTRONICS AND TELECOMMUNICATIONS RESEARCH INSTITUTE)

From WikiPatents
Jump to navigation Jump to search

APPARATUS AND METHOD FOR IMAGE OUTPAINTING BASED ON LEARNING OF A SINGLE INPUT IMAGE

Organization Name

ELECTRONICS AND TELECOMMUNICATIONS RESEARCH INSTITUTE

Inventor(s)

Jung-Jae Yu of Daejeon (KR)

Ju-Won Lee of Daejeon (KR)

APPARATUS AND METHOD FOR IMAGE OUTPAINTING BASED ON LEARNING OF A SINGLE INPUT IMAGE - A simplified explanation of the abstract

This abstract first appeared for US patent application 20230169628 titled 'APPARATUS AND METHOD FOR IMAGE OUTPAINTING BASED ON LEARNING OF A SINGLE INPUT IMAGE

Simplified Explanation

The patent application describes an apparatus and method for image outpainting using a single input image. The apparatus generates multiple extended images by applying different noise vectors to the input image. It then trains a discriminator network to minimize the loss of a generative adversarial network using the first extended image, the input image, and a horizontally flipped version of the input image.

  • The apparatus generates multiple extended images from a single input image.
  • It uses different noise vectors to create variations in the extended images.
  • The discriminator network is trained to minimize the loss of a generative adversarial network.
  • The input image, first extended image, and horizontally flipped image are used for training the discriminator network.
  • The apparatus also learns the multiple extended images by minimizing the sum of two differences.
  • The first difference is between the input image and the center region of a second extended image.
  • The second difference is between vertically segmented and flipped segments of the input image and the left and right regions of a third extended image.

Potential applications of this technology:

  • Image editing and manipulation software
  • Virtual reality and augmented reality applications
  • Artistic and creative tools for generating new images

Problems solved by this technology:

  • Limited availability of high-quality training data for image outpainting
  • Difficulty in generating realistic and diverse extended images from a single input image

Benefits of this technology:

  • Enables the generation of multiple extended images from a single input image
  • Provides a method for training a discriminator network to improve the quality of generated images
  • Allows for the creation of realistic and diverse extended images for various applications


Original Abstract Submitted

disclosed herein are an apparatus and method for image outpainting based on learning of a single input image. the apparatus generates multiple extended images from an input image using multiple noise vectors, inputs a first extended image, among the multiple extended images, the input image, and an image acquired by horizontally flipping the input image and thereby trains a discriminator network such that the loss of a generative adversarial network is minimized, and learns the multiple extended images such that the sum of two differences is minimized, the first one being the difference between the input image and the center region of a second extended image, and the second one being the difference between images acquired by vertically segmenting the input image and flipping the segments and the left and right regions of a third extended image.

Retrieved from "http://wikipatents.org/index.php?title=20230169628._APPARATUS_AND_METHOD_FOR_IMAGE_OUTPAINTING_BASED_ON_LEARNING_OF_A_SINGLE_INPUT_IMAGE_simplified_abstract_(ELECTRONICS_AND_TELECOMMUNICATIONS_RESEARCH_INSTITUTE)&oldid=25431"