Virtual Viewpoint Image Generation with Color Consistency

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image processing techniques for generating virtual viewpoint images from multi-viewpoint images often result in unnatural color changes around the boundaries of areas textured-mapped from different image capturing apparatuses, causing visual oddness due to lighting and equipment differences.

Innovation Solution

An image processing apparatus that determines pixel values for virtual viewpoint images by combining pixel values from different image capturing apparatuses, using a determination unit to calculate pixel values for areas not captured by one apparatus at a specific time based on values from other apparatuses, and generates virtual viewpoint images by transforming and combining background, shadow, and object image data to minimize color discrepancies.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Area of stationary object

If texture images from multiple image capturing apparatuses are blended and texture-mapped to generate a virtual viewpoint image, then the coverage and completeness of the virtual viewpoint image are improved, but unnatural color changes appear around the boundaries between areas from different apparatuses

Engineering Contradiction:
Improvecoverage area of virtual viewpoint imageVSAvoidcolor consistency at boundaries
Core Design Contradiction:
Area of stationary objectVSManufacturing precision

Solution Approach 1:

The patent applies local quality by determining pixel values differently based on local conditions: for pixels in areas captured by the first image capturing apparatus, pixel values are determined based on temporal changes (first time and second time images), while for pixels in areas not captured by the first apparatus, pixel values are determined based on the second image capturing apparatus. This localized approach ensures color consistency within each region while maintaining overall image completeness.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent uses an intermediary approach by introducing a determination unit that selectively chooses which image capturing apparatus's data to use for each pixel. This intermediary decision-making process prevents direct blending of color-different images at boundaries, instead using temporal data from the same apparatus or spatial data from a different apparatus only when necessary, thereby maintaining color consistency.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Adaptability or versatility

If images from different image capturing apparatuses are combined to create a virtual viewpoint image, then the versatility and viewpoint flexibility are improved, but visual oddness occurs due to lighting and equipment differences between apparatuses

Engineering Contradiction:
Improvevirtual viewpoint flexibilityVSAvoidvisual coherence of generated image
Core Design Contradiction:
Adaptability or versatilityVSReliability

Solution Approach 1:

The patent ensures visual coherence by applying local quality: each pixel's value is determined based on the specific capturing conditions and temporal availability. Pixels are processed differently depending on which apparatus captured them and when, ensuring that lighting and equipment characteristics remain consistent within each region while still achieving versatile viewpoint generation.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent applies preliminary action by pre-determining which pixels should be filled from which image capturing apparatus based on temporal and spatial analysis. The determination unit proactively identifies areas not captured by the first apparatus and prepares to fill them from the second apparatus, preventing visual oddness before it occurs in the final image generation.

Inventive Principle:
Principle #10Preliminary action

3Measurement precision

If pixel values are determined using images from the first image capturing apparatus at different times, then temporal resolution and motion accuracy are improved, but the completeness of the virtual viewpoint image decreases for areas not captured at the first time

Engineering Contradiction:
Improvetemporal precision of pixel valuesVSAvoidmissing pixel data in uncaptured areas
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The patent merges data from multiple image capturing apparatuses and multiple time points to create a complete virtual viewpoint image. The determination unit combines temporal data (first time and second time images from the first apparatus) with spatial data (images from the second apparatus) to fill gaps, ensuring both temporal precision and information completeness are achieved.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The determination unit performs preliminary analysis to identify which pixels are available from temporal data and which require spatial data from other apparatuses. This preliminary action ensures that temporal precision is maintained where possible, while information completeness is restored by selectively incorporating data from the second image capturing apparatus.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentUS10728513B2Image processing apparatus, image processing method, and storage medium
Publication Date: 2020.07.28 CANON KK
  • US10728513B2 patent drawing
  • US10728513B2 patent drawing
  • US10728513B2 patent drawing

AI summary

An image processing apparatus includes: an image obtaining unit configured to obtain images of an image capturing target area from a plurality of directions; an information obtaining unit configured to obtain viewpoint information; a determination unit configured to determine a pixel value of a particular area in a virtual viewpoint image according to a virtual viewpoint at a first time indicated by the viewpoint information, based on a pixel value of an obtained image based on image capturing by a first image capturing apparatus at a second time, and a pixel value of an obtained image based on image capturing by a second image capturing apparatus at the first time, wherein the particular area is not captured by the first image capturing apparatus at the first time; and a generation unit configured to generate a virtual viewpoint image in accordance with the determination of the pixel value.