Dead Leaves Image Synthesis with Texture Blending for AI Data

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing dead leaves models for synthetic data generation have limitations, such as only modeling circular discs, limited color sampling, and equal weighting of texture and background color, which restrict their flexibility and realism.

Innovation Solution

The proposed system and method generate synthetic data using dead leaves images by obtaining a raw image, determining its color distribution and variation, and using an iterative process to create a dead leaf image by adding circles and sticks, while blending textures based on the image's color distribution and variation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional dead leaves models are used for synthetic data generation, then the generation process is simple, but the realism and flexibility of generated images are limited

Engineering Contradiction:

Improverealism of synthetic imagesVSAvoidcomplexity of generation process

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The generation process is divided into distinct stages: initial dead leaves image generation using traditional models, followed by separate texture synthesis and blending stages. This segmentation allows each stage to be optimized independently, improving overall realism without requiring complete process redesign.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

Color distribution and variation parameters are pre-calculated from reference images before the main generation process. This preliminary action enables the iterative blending process to efficiently achieve realistic results by using pre-computed statistical properties as guides.

Inventive Principle:

Principle #10Preliminary action

2Adaptability or versatility

If only circular discs are modeled in dead leaves models, then the model structure is simple, but the versatility and realism of generated images are restricted

Engineering Contradiction:

Improvevariety of shapes in generated imagesVSAvoidcomplexity of model structure

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The model transitions from symmetric circular discs to asymmetric irregular shapes that better represent natural objects. The iterative blending process incorporates shape variations while maintaining computational efficiency, enabling diverse shape generation without proportionally increasing complexity.

Inventive Principle:

Principle #4Asymmetry

Solution Approach 2:

The model structure becomes dynamic and adaptable through iterative refinement. Instead of being fixed to circular discs, the model can adapt shapes and textures through multiple blending iterations, allowing versatility to increase with controlled complexity growth.

Inventive Principle:

Principle #15Dynamics

3Manufacturing precision

If equal weighting is applied to texture and background color, then the processing is straightforward, but the quality and realism of texture modeling deteriorates

Engineering Contradiction:

Improvequality of texture modelingVSAvoidcomplexity of blending process

Core Design Contradiction:

Manufacturing precisionVSDevice complexity

Solution Approach 1:

Different regions of the generated image receive different weighting during texture blending based on local characteristics. Areas requiring detailed texture representation receive higher texture weights, while other regions maintain appropriate background color prominence. This local adaptation improves texture modeling quality without requiring uniformly complex processing throughout.

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The blending weights for texture and background color are dynamically adjusted as parameters during the iterative process. These parameter changes enable precise control over texture modeling quality, allowing the system to optimize realism by varying weights based on image content and desired outcomes.

Inventive Principle:

Principle #35Parameter changes

4Productivity

If ground-truth data is used for training AI models, then the training accuracy is high, but data availability and generation time increase

Engineering Contradiction:

Improvespeed of data generationVSAvoidquality of training data

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The system creates synthetic copies of realistic images through the enhanced dead leaves model rather than using actual ground-truth photographs. By copying the statistical properties and visual characteristics of real images through iterative synthesis, the system achieves rapid data generation while maintaining training quality suitable for AI model development.

Inventive Principle:

Principle #26Copying

Data Source

PatentUS12347110B2System and method for synthetic data generation using dead leaves images

Publication Date: 2025.07.01 SAMSUNG ELECTRONICS CO LTD

AI summary

A method includes obtaining a raw image in a first image domain. The method also includes determining a color distribution and an amount of variation in the raw image. The method further includes using an iterative process to generate a dead leaf image from a blank image in the first image domain. The iterative process includes adding multiple circles and multiple sticks to the blank image until the dead leaf image is filled. The iterative process also includes blurring portions of the dead leaf image during at least one iteration of the iterative process. Textures of the multiple circles are blended based on the color distribution and the amount of variation in the raw image.