Log-Chromaticity Clustering for Illumination Separation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current computer vision applications face challenges in accurately separating illumination and material aspects of images, which affects the accuracy of object recognition and optical character recognition.

Innovation Solution

A method and system that generate intrinsic images by creating a bi-illuminant chromaticity plane in log color space, providing a set of estimates for orientation, and generating normal maps to accurately represent image characteristics, using techniques such as log-chromaticity clustering and bi-illuminant dichromatic reflection models to distinguish between illumination and material.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional image processing techniques are used, then the processing is simpler, but the accuracy of separating illumination and material aspects is insufficient

Engineering Contradiction:

Improveaccuracy of separating illumination and materialVSAvoidcomplexity of image processing system

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the image processing into distinct modules: chromaticity plane generation, normal map generation, and intrinsic image generation. Each module handles a specific aspect of the separation task, allowing the system to achieve high accuracy through structured decomposition of the complex problem into manageable components

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent transforms the image data from RGB color space into log-chromaticity space by projecting onto a bi-illuminant chromaticity plane. This dimensional transformation enables better separation of illumination and material characteristics by reorganizing the color information in a different mathematical space that highlights the desired separability

Inventive Principle:

Principle #17Another dimension (Dimensionality change)

2Manufacturing precision

If more sophisticated image processing techniques are applied, then the accuracy and precision of intrinsic image generation improve, but the processing time increases

Engineering Contradiction:

Improveprecision of intrinsic image generationVSAvoidimage processing time

Core Design Contradiction:

Manufacturing precisionVSLoss of time

Solution Approach 1:

The patent performs preliminary computations by pre-calculating the bi-illuminant chromaticity plane and generating normal maps before the final intrinsic image generation. These pre-computed structures serve as foundations that accelerate the subsequent separation process, reducing the computational burden during the critical separation step

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent replaces iterative and computationally intensive optimization methods with direct analytical projections onto the chromaticity plane. By using mathematical projections and pre-computed normal maps, the system achieves high precision without requiring time-consuming iterative adjustments or complex optimization algorithms

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS8849018B2Log-chromaticity clustering pipeline for use in an image process

Publication Date: 2014.09.30 INNOVATION ASSET COLLECTIVE

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 defined by image locations, in a computer memory, generating a bi-illuminant chromaticity plane in a log color space for representing the image locations of the image in a log-chromaticity representation for the image, providing a set of estimates for the orientation of the bi-illuminant chromaticity plane and calculating an orientation for each one of the image locations as a function of the set of estimates for the orientation.