Image Reflection Separation via Frequency Analysis

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for separating diffusive and specular reflection components from a single image suffer from low accuracy, especially at dichromatic boundaries and pixels with zero chroma, leading to degraded image quality due to artifacts when processed further.

Innovation Solution

An image processing apparatus and method that obtains diffusive and specular reflection components from a single image with high accuracy by performing filter-processing and evaluating separation accuracy based on frequency characteristics, using a modification value obtained through an energy function to improve separation accuracy pixel by pixel.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If diffusive and specular reflection components are separated from a single image using noise model or bilateral filtering, then processing complexity is reduced, but separation accuracy deteriorates especially at dichromatic boundaries and pixels with zero chroma

Engineering Contradiction:
Improveprocessing complexityVSAvoidseparation accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent segments the separation process into multiple passes: first separating components from the original image, then identifying low-accuracy regions through frequency analysis, and finally applying targeted filtering only to those regions. This segmentation allows the system to maintain low overall complexity while achieving high accuracy where needed.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies different processing strategies to different regions of the image. Frequency analysis identifies specific regions with low separation accuracy (such as dichromatic boundaries), and filtering is applied locally to those regions rather than uniformly across the entire image, optimizing both accuracy and efficiency.

Inventive Principle:
Principle #3Local quality

2Manufacturing precision

If filter-processing is applied to improve image quality, then visual effects are enhanced, but artifacts are generated due to low separation accuracy

Engineering Contradiction:
Improveimage qualityVSAvoidartifacts
Core Design Contradiction:
Manufacturing precisionVSObject-generated harmful factors

Solution Approach 1:

The patent performs frequency analysis and identifies low-accuracy regions before applying filter-processing. By preliminarily determining which regions need improvement, the system avoids applying filters to already accurate regions, thereby preventing artifact generation while still enhancing image quality where needed.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent uses frequency characteristics as feedback to evaluate separation accuracy and guide subsequent filtering operations. The frequency analysis provides information about where separation accuracy is low, and this feedback directs the filtering process to those specific regions, improving image quality without generating artifacts in accurate regions.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP3488412B1Image processing apparatus, image processing method, and computer-readable recording medium thereof
Publication Date: 2020.08.12 SAMSUNG ELECTRONICS CO LTD
  • EP3488412B1 patent drawingFigure 1
  • EP3488412B1 patent drawingFigure 2
  • EP3488412B1 patent drawingFigure 3

AI summary

Provided is an image processing apparatus that includes a component obtainer configured to obtain diffusive reflection components and specular reflection components for pixels of an input image, a filter processor configured to perform filter-processing on a diffusive reflection component image, a specular reflection component image, and the input image, a component combiner configured to combine the filter-processed diffusive reflection component image and the filter-processed specular reflection component image to generate a combined image, and an evaluator configured to evaluate a separation accuracy of the diffusive reflection components or the specular reflection components based on the combined image and the filter-processed input image.