公开(公告)号：US20200311870A1

公开(公告)日：2020-10-01

申请号：US16831805

申请日：2020-03-27

Applicant: AGENCY FOR DEFENSE DEVELOPMENT ,  KOREA ADVANCED INSTITUTE OF SCIENCE AND TECHNOLOGY

Inventor： Ju Hyun JUNG ,  Dong Hyun KIM ,  No Hyeok PARK ,  Jeung Won CHOI ,  Dae Eun KIM ,  Se Hwan KI ,  Mun Churl KIM ,  Ki Nam JUN ,  Seung Ho BAEK ,  Jong Hwan KO

IPC: G06T3/40 ,  H04N19/59 ,  H04N19/33 ,  G06T9/00 ,  G06N3/04 ,  G06N3/08

Abstract: Provided are a method and an apparatus for performing scalable video decoding, wherein the method and the apparatus down-sample input video, determine the down-sampled input video as base layer video, generate prediction video for enhancement layer video by applying an up-scaling filter to the base layer video, and code the base layer video and the prediction video, wherein the up-scaling filter is a convolution filter of a deep neural network.

公开(公告)号：US20220012855A1

公开(公告)日：2022-01-13

申请号：US17482433

申请日：2021-09-23

Applicant: KOREA ADVANCED INSTITUTE OF SCIENCE AND TECHNOLOGY

Inventor： Mun Churl KIM ,  Soo Ye KIM ,  Dae Eun KIM

IPC: G06T5/00 ,  G06F17/11 ,  G06N3/02 ,  G06N3/08 ,  G06T5/50

Abstract: In this invention, we propose a convolutional neural network (CNN) based architecture designed for the ITM to HDR consumer displays, called ITM-CNN, and its training strategy for enhancing the performance based on image decomposition using the guided filter. We demonstrate the benefits of decomposing the image by experimenting with various architectures and also compare the performance for different training strategies. To the best of our knowledge, this invention first presents the ITM problem using CNNs for HDR consumer displays, where the network is trained to restore lost details and local contrast. Our ITM-CNN can readily up-convert LDR images for direct viewing on an HDR consumer medium, and is a very powerful means to solve the lack of HDR video contents with legacy LDR videos.

公开(公告)号：US20170372461A1

公开(公告)日：2017-12-28

申请号：US15633763

申请日：2017-06-27

Applicant: SILICON WORKS CO., LTD. ,  Korea Advanced Institute of Science and Technology

Inventor： Yong Woo KIM ,  Sang Yeon KIM ,  Woo Suk HA ,  Mun Churl KIM ,  Dae Eun KIM

IPC: G06T5/00 ,  G06T5/10 ,  G06T7/90 ,  G06T5/20

Abstract: The present invention provides a technology that separates a low-contrast-ratio image into sublayer images, classifies each sublayer image into several categories in accordance with the characteristics of each sublayer image, and learns a transformation matrix representing a relationship between the low-contrast-ratio image and a high-contrast-ratio image for each category. In addition, the present invention provides a technology that separates an input low-contrast-ratio image into sublayer images, selects a category corresponding to each sublayer image, and applies a learned transformation matrix to generate a high.

公开(公告)号：US20210166360A1

公开(公告)日：2021-06-03

申请号：US16769576

申请日：2017-12-06

Applicant: KOREA ADVANCED INSTITUTE OF SCIENCE AND TECHNOLOGY

Inventor： Mun Churl KIM ,  Soo Ye KIM ,  Dae Eun KIM

IPC: G06T5/00 ,  G06T5/50

Abstract: Inverse tone mapping (ITM) aims at generating a single high dynamic range (HDR) image from a low dynamic range (LDR) image. While ITM was frequently used for graphics rendering in the HDR space, the advent of HDR consumer displays (e.g., HDR TV) and the consequent need for HDR multimedia contents open up new horizons for the consumption of ultra-high quality video contents. However, due to the lack of HDR-filmed contents, the legacy LDR videos must be up-converted for viewing on these HDR displays. Unfortunately, the previous ITM methods are not appropriate for HDR consumer displays, and their inverse-tone-mapped results are not visually pleasing with noise amplification or lack of details. In this paper, we propose a convolutional neural network (CNN) based architecture designed for the ITM to HDR consumer displays, called ITM-CNN, and its training strategy for enhancing the performance based on image decomposition using the guided filter. We demonstrate the benefits of decomposing the image by experimenting with various architectures and also compare the performance for different training strategies. To the best of our knowledge, this paper first presents the ITM problem using CNNs for HDR consumer displays, where the network is trained to restore lost details and local contrast. Our ITM-CNN can readily up-convert LDR images for direct viewing on an HDR consumer medium, and is a very powerful means to solve the lack of HDR video contents with legacy LDR videos.