Automated Image Coloring via Neural Network Mask Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The rapid growth of the online content market and increasing user demand for webtoon content have outpaced traditional content creation methods, leading to inefficiencies in producing colored content, such as webtoons, which require significant time and effort from creators.
Innovation Solution
A method utilizing trained artificial neural networks to automatically color target images by segmenting and matching them to reference images, reducing the time and effort required for coloring by generating target masks, reference masks, and colored target images, while ensuring color consistency and applicability of image effects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional manual coloring methods are used, then color quality and artistic control are maintained, but content production speed is slow and cannot keep up with market expansion
Solution Approach 1:
The patent replaces the manual mechanical coloring process with an automated image processing system that uses reference images and algorithmic color transfer. The system automatically analyzes reference images to extract color information and applies it to target images, eliminating the need for manual coloring while maintaining color quality and consistency.
Solution Approach 2:
The patent uses reference images as templates to copy color information onto target images. By analyzing the color patterns, gradients, and styling in reference images, the system replicates these visual characteristics automatically, enabling rapid production of colored content that matches the desired artistic style.
2Productivity
If automatic coloring methods are implemented, then coloring speed is improved, but color quality and suitability to user needs may deteriorate
Solution Approach 1:
The patent incorporates feedback mechanisms where the system continuously compares the generated colored images against reference images to ensure color consistency and quality. The algorithm adjusts color parameters based on the visual characteristics of reference images, providing automatic quality control that maintains color suitability while enabling rapid production.
Solution Approach 2:
The patent dynamically adjusts color parameters such as hue, saturation, and brightness based on the analysis of reference images. By changing these parameters automatically according to the reference image characteristics, the system maintains high color quality and artistic suitability while achieving fast coloring speeds through algorithmic optimization.
Data Source
AI summary
A method of coloring a target image includes generating at least one target mask including a part of the target image by using a first artificial neural network, which has been trained; generating at least one reference mask that corresponds to the at least one target mask and includes at least part of a reference image, by using the first artificial neural network; generating at least one colored target mask by coloring the at least one target mask in reference to the color of the at least one reference mask; and generating a colored target image from the target image, the at least one target mask, and the at least one colored target mask by using a second artificial neural network, which has been trained.