Inter-view prediction areas in multi-camera 360-degree imaging

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current multiview video coding systems are inefficient in encoding and transmitting 360-degree content due to high redundancy and data intensity, particularly in applications like Head Mounted Displays and virtual reality, where current broadcasting infrastructure struggles to handle the large data requirements in real-time.

Innovation Solution

A Multiview camera device with multiple cameras captures a 360-degree view, using a controller and memory to determine areas for inter-view prediction, enabling efficient data compression by utilizing the overlap between camera views, thereby reducing the bitrate required for transmission.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of substance

If inter-view prediction is enabled over the entire image, then data compression is improved, but image quality deteriorates in non-overlapping areas

Engineering Contradiction:

Improvedata redundancyVSAvoidimage quality

Core Design Contradiction:

Loss of substanceVSManufacturing precision

Solution Approach 1:

The patent applies local quality by enabling inter-view prediction only in specific overlapping areas between adjacent camera views rather than uniformly across the entire image. The system determines overlap regions based on camera positioning and field-of-view characteristics, then selectively applies prediction algorithms only where redundancy exists, preserving image quality in non-overlapping areas while achieving compression in overlapping regions.

Inventive Principle:

Principle #3Local quality

2Adaptability or versatility

If multiple cameras are used to capture 360-degree content, then viewing coverage is improved, but data transmission requirements worsen

Engineering Contradiction:

Improveviewing coverageVSAvoiddata volume

Core Design Contradiction:

Adaptability or versatilityVSQuantity of substance

Solution Approach 1:

The patent merges adjacent camera views in overlapping areas by using inter-view prediction to combine information from multiple cameras. Instead of transmitting separate full-resolution images from each camera, the system uses one view as a reference and predicts the other view in overlapping regions, significantly reducing the total data volume while maintaining 360-degree viewing coverage.

Inventive Principle:

Principle #5Merging (Combining)

3Loss of substance

If inter-view prediction is applied in non-overlapping areas, then compression is improved, but prediction accuracy worsens

Engineering Contradiction:

Improvedata redundancyVSAvoidprediction accuracy

Core Design Contradiction:

Loss of substanceVSMeasurement precision

Solution Approach 1:

The patent applies partial action by limiting inter-view prediction to only the overlapping areas between camera views. The system identifies boundary regions where overlap exists and applies prediction algorithms only within those boundaries, avoiding application in non-overlapping areas where it would reduce accuracy. This selective approach ensures compression is achieved where beneficial without compromising prediction accuracy.

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentEP3507986B1Determining inter-view prediction areas in images captured with a multi-camera device

Publication Date: 2023.12.13 NOKIA TECHNOLOGIES OY

Figure 1
Figure 2
Figure 3

AI summary

An apparatus comprising a multi-camera device including at least two cameras and a controller for receiving image data from the cameras, determining which areas of the images to enable for inter-view prediction, wherein the determination is based on imaging characteristics of the cameras and relative locations and orientations of the cameras within the multi-camera device, and enabling inter-view prediction for the determined areas of the images.