Processing

Please wait...

Settings

Settings

Goto Application

1. JP2020107239 - IMAGE PROCESSING METHOD, IMAGE PROCESSING SYSTEM AND PROGRAM

Office
Japan
Application Number 2018247800
Application Date 28.12.2018
Publication Number 2020107239
Publication Date 16.03.2020
Grant Number 6676744
Grant Date 16.03.2020
Publication Kind B1
IPC
G06T 11/80
GPHYSICS
06COMPUTING; CALCULATING OR COUNTING
TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
112D image generation
80Creating or modifying a manually drawn or painted image using a manual input device, e.g. mouse, light pen, direction keys on keyboard
G06N 20/00
GPHYSICS
06COMPUTING; CALCULATING OR COUNTING
NCOMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
20Machine learning
G06T 1/00
GPHYSICS
06COMPUTING; CALCULATING OR COUNTING
TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
1General purpose image data processing
G06T 1/40
GPHYSICS
06COMPUTING; CALCULATING OR COUNTING
TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
1General purpose image data processing
20Processor architectures; Processor configuration, e.g. pipelining
40Neural networks
CPC
G06T 1/00
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
1General purpose image data processing
G06T 11/80
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
112D [Two Dimensional] image generation
80Creating or modifying a manually drawn or painted image using a manual input device, e.g. mouse, light pen, direction keys on keyboard
G06N 20/00
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
NCOMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
20Machine learning
G06N 3/049
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
NCOMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
3Computer systems based on biological models
02using neural network models
04Architectures, e.g. interconnection topology
049Temporal neural nets, e.g. delay elements, oscillating neurons, pulsed inputs
G06T 11/001
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
112D [Two Dimensional] image generation
001Texturing; Colouring; Generation of texture or colour
G06T 11/203
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
112D [Two Dimensional] image generation
20Drawing from basic elements, e.g. lines or circles
203Drawing of straight lines or curves
Applicants CYGAMES CO LTD
株式会社Cygames
Inventors KURABAYASHI SHUICHI
倉林 修一
Agents 速水 進治
天城 聡
Title
(EN) IMAGE PROCESSING METHOD, IMAGE PROCESSING SYSTEM AND PROGRAM
(JA) 画像処理方法、画像処理システム及びプログラム
Abstract
(EN)

PROBLEM TO BE SOLVED: To color, with high accuracy, line drawings of characters of a game or an animation.

SOLUTION: An image processing method includes: storing learning data in which a line drawing image and a colored image are paired for each element; generating a model for use in estimating the colored image from the line drawing image by machine learning using the learning data; acquiring the line drawing image to be colored; generating the colored image for combination with a target line drawing image based on an estimation model corresponding to a specific element of the target line drawing image and the target line drawing image; generating a colored layer of an image file including a line drawing layer and a colored layer using a target colored image; inserting the generated colored layer into the image file including the line drawing layer corresponding to the target line drawing image; correcting, based on a user input, the colored layer of the target image file into which the colored layer is inserted; extracting the corrected colored layer and the corresponding line drawing layer as a pair of images for learning; and storing the pair of the line drawing image of the extracted line drawing layer and the colored image of the extracted colored layer in association with an element corresponding to the estimation model used upon generation of the target colored image.

SELECTED DRAWING: Figure 9

COPYRIGHT: (C)2020,JPO&INPIT


(JA)

【課題】ゲームやアニメのキャラクタ線画を高精度に着色する。
【解決手段】画像処理方法は、要素ごとに線画画像及び着色画像をペアとする学習データを記憶し、学習データを用いた機械学習により、線画画像から着色画像を推定するモデルを生成し、着色対象の線画画像を取得し、対象線画画像の特定要素に対応する推定モデルと対象線画画像に基づき対象線画画像に組み合わせる着色画像を生成し、対象着色画像を用いて線画レイヤ及び着色レイヤを含む画像ファイルの着色レイヤを生成し、対象線画画像に対応する線画レイヤを含む画像ファイルに挿入し、ユーザ入力に基づき、着色レイヤを挿入された対象画像ファイルの着色レイヤを修正し、修正した着色レイヤと対応する線画レイヤとを学習用の画像ペアとして抽出し、抽出した線画レイヤの線画画像及び着色レイヤの着色画像のペアを、対象着色画像を生成する時に用いた推定モデルに対応する要素に対応付けて記憶する。
【選択図】図9


Related patent documents