Intrinsic Image Post-Processing Using Bi-Illuminant Dichromatic Model

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current computer vision applications face challenges in accurately separating illumination and material aspects of images, which affects the accuracy of image processing and analysis.

Innovation Solution

A method and system using a bi-illuminant, dichromatic reflection model to generate intrinsic images by processing image files, identifying token regions, and applying post-processing techniques such as edge-preserving blur and artifact reduction to enhance image separation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If current image processing methods are used to separate illumination and material aspects, then the basic separation can be achieved, but the accuracy of the separation is insufficient and artifacts remain

Engineering Contradiction:
Improveaccuracy of illumination and material separationVSAvoidartifacts in intrinsic images
Core Design Contradiction:
Measurement precisionVSObject-generated harmful factors

Solution Approach 1:

The patent introduces a post-processing step that uses the bi-illuminant, dichromatic reflection model as an intermediary to refine the intrinsic images. This intermediary process corrects artifacts by enforcing physical constraints on illumination and material separation, thereby improving accuracy without directly modifying the original separation algorithm

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent implements a feedback mechanism where the generated intrinsic images are evaluated against the bi-illuminant, dichromatic reflection model, and corrections are applied based on the model's constraints. This feedback loop continuously refines the separation accuracy by comparing results with physical expectations and adjusting accordingly

Inventive Principle:
Principle #23Feedback

2Measurement precision

If simple intrinsic image generation is used, then processing speed is maintained, but accuracy and quality of image separation deteriorates

Engineering Contradiction:
Improveaccuracy of intrinsic image generationVSAvoidcomplexity of processing steps
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent performs preliminary action by first generating intrinsic images using existing methods, then applying post-processing corrections. This two-stage approach allows the system to benefit from simple initial processing while adding complexity only where needed to improve accuracy, rather than requiring complex processing throughout

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent applies local quality by using the bi-illuminant, dichromatic reflection model to selectively correct specific regions and aspects of the intrinsic images where artifacts are most problematic, rather than uniformly processing the entire image. This targeted approach improves accuracy in critical areas while maintaining overall processing efficiency

Inventive Principle:
Principle #3Local quality

Data Source

PatentEP2776979B1Post processing for improved generation of intrinsic images
Publication Date: 2018.12.26 TANDENT VISION SCIENCE INC
  • EP2776979B1 patent drawingFigure 1
  • EP2776979B1 patent drawingFigure 2
  • EP2776979B1 patent drawingFigure 3A

AI summary

In a first exemplary embodiment of the present invention, an automated, computerized method is provided for processing an image. According to a feature of the present invention, the method comprises the steps of providing an image file depicting an image, in a computer memory, generating an intrinsic image corresponding to the image and modifying the intrinsic image as a function of a bi-illuminant, dichromatic reflection model.