Retinal Noise Texture for Digital Images

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for adding texture to digital images, such as film grain emulation, are computationally intensive, restrict creative freedom, and increase bit rates, making them inefficient for real-time implementation and compression in high-resolution video formats.

Innovation Solution

A computer-implemented method that transforms a digital image to emulate retinal noise by applying a nonlinear photoreceptor response and a center-surround kernel, followed by adding noise with a bandpass frequency spectrum, and then inverting the transformation to achieve a natural, low-complexity texture addition.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If film grain emulation methods are used to add texture to digital images, then the visual appearance and realism of the image is improved, but the computational complexity increases significantly

Engineering Contradiction:
Improvevisual appearance qualityVSAvoidcomputational complexity
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent applies the inversion principle by transforming the image through a nonlinear photoreceptor response and center-surround kernel to simulate retinal noise, then inverting the transformation to add texture. This approach reverses the conventional method of directly overlaying grain and achieves realistic texture with lower computational complexity by modeling the biological noise generation process rather than simply superimposing pre-computed grain patterns.

Inventive Principle:
Principle #13The other way round (Inversion)

Solution Approach 2:

The patent changes the approach from using fixed film grain patterns to dynamically modeling retinal noise through parameterizable transformations. The nonlinear photoreceptor response and center-surround kernel allow for flexible control of noise characteristics, enabling adaptation to different image content and viewing conditions while maintaining computational efficiency through analytical inversion formulas.

Inventive Principle:
Principle #35Parameter changes

2Manufacturing precision

If existing film grain emulation methods are applied to digital images, then texture is added to improve appearance, but the bit rate increases requiring higher compression resources

Engineering Contradiction:
Improveimage qualityVSAvoidbit rate
Core Design Contradiction:
Manufacturing precisionVSQuantity of substance

Solution Approach 1:

The patent converts the harmful effect of compression artifacts into a beneficial outcome by adding retinal noise that masks these artifacts. The noise generated through the retinal model serves dual purposes: it provides the desired texture improvement while simultaneously hiding compression artifacts, thereby achieving high image quality at lower bit rates compared to conventional methods that treat texture addition and artifact masking as separate operations.

Inventive Principle:
Principle #22Blessing in disguise (Convert harm into benefit)

3Manufacturing precision

If computationally intensive texture addition methods are used, then realistic film grain effect is achieved, but real-time implementation is prevented

Engineering Contradiction:
Improvetexture realismVSAvoidreal-time processing capability
Core Design Contradiction:
Manufacturing precisionVSProductivity

Solution Approach 1:

The patent enables real-time processing by inverting the transformation sequence. Instead of applying complex grain overlay operations that are computationally intensive in the forward direction, the method uses an inverted approach where the retinal noise is generated and added in a transformed domain, allowing for efficient real-time implementation while maintaining realistic texture appearance.

Inventive Principle:
Principle #13The other way round (Inversion)

Data Source

PatentEP3866116B1Computer-implemented method for adding texture to a digital image
Publication Date: 2023.08.23 UNIV POMPEU FABRA

AI summary

A computer-implemented method for adding texture to a digital image, wherein the computer has a digital image, comprising the method the following steps: - transforming the digital image into an image R that emulates the retinal output; - adding noise nr to the image R in order to obtain a noisy image Rr; - applying to the noisy image Rr the inverse of the transformation applied to the digital image in order to obtain a final image O.