IMAGE QUALITY IMPROVEMENT APPARATUS USING ARTIFICIAL NEURAL NETWORK
The present invention is to provide an image with improved image quality using an artificial neural network. An image quality improvement device includes: a model generation unit for generating an adversarial generative neural network model including a generator for generating a similar high-definit...
Gespeichert in:
Hauptverfasser: | , |
---|---|
Format: | Patent |
Sprache: | eng ; kor |
Schlagworte: | |
Online-Zugang: | Volltext bestellen |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | The present invention is to provide an image with improved image quality using an artificial neural network. An image quality improvement device includes: a model generation unit for generating an adversarial generative neural network model including a generator for generating a similar high-definition sample from a first image and a discriminator for determining the authenticity of the similar high-definition sample; a learning unit that trains the adversarial generative neural network model which allows the generator to extract a sample of at least a portion of the first image, downscale the sample, and improve quality of the downscaled sample to generate a similar high-definition sample, and allows the discriminator to determine authenticity of the similar high-definition sample; and an image quality improvement unit that improves quality of a second image by inputting the second image to the trained adversarial generative neural network model.
이미지 화질 개선 장치는 제 1 이미지로부터 유사 고화질 샘플을 생성하는 생성자 및 유사 고화질 샘플의 진위 여부를 판별하는 판별자를 포함하는 적대적 생성 신경망 모델을 생성하는 모델 생성부, 생성자가 제 1 이미지의 적어도 일부의 샘플을 추출하고, 샘플을 다운 스케일링하고, 다운 스케일링한 샘플의 화질을 개선하여 유사 고화질 샘플을 생성하고, 판별자가 유사 고화질 샘플의 진위 여부를 판별하도록 적대적 생성 신경망 모델을 학습시키는 학습부 및 학습된 적대적 생성 신경망 모델에 제 2 이미지를 입력하여 제 2 이미지의 화질을 개선하는 이미지 화질 개선부를 포함할 수 있다. |
---|