Non-Uniform Image Projection for Resolution Preservation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional image processing methods for panoramic images with non-uniform information density, such as those captured by fisheye lenses, result in loss of image resolution due to uniform downsampling and encoding, which is not compatible with existing hardware and software ecosystems, leading to suboptimal user experience and increased costs with higher-level codecs.

Innovation Solution

The method involves generating an unequal area (UA) or non-uniform (NU) projection from captured image content, which is then encoded and transmitted in segments, preserving source lens information and optimizing for hardware-based processing, using techniques like unequal area cubemap (UAC) projections and lossless transforms to fit within existing codec limitations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If uniform downsampling and encoding is applied to panoramic images with non-uniform information density, then compatibility with existing hardware and software ecosystems is achieved, but image resolution is lost

Engineering Contradiction:

Improvecompatibility with existing hardware and software ecosystemsVSAvoidimage resolution

Core Design Contradiction:

Adaptability or versatilityVSManufacturing precision

Solution Approach 1:

The patent applies non-uniform downsampling where different regions of the image are processed differently based on their information density. Central regions with higher information density retain more pixels, while peripheral regions with lower information density are downsampled more aggressively. This local differentiation resolves the contradiction by preserving image resolution in critical areas while maintaining compatibility through overall downsampling.

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The patent changes the downsampling parameter from uniform to non-uniform based on spatial position and information density. By varying the downsampling ratio across different regions of the image, the system achieves both compatibility (through downsampling) and resolution preservation (through selective retention of high-density regions).

Inventive Principle:

Principle #35Parameter changes

2Device complexity

If uniform scaling downsampling is applied over the entire image, then processing simplicity is maintained, but image resolution is lost and information density distribution degrades

Engineering Contradiction:

Improveprocessing simplicityVSAvoidimage resolution

Core Design Contradiction:

Device complexityVSManufacturing precision

Solution Approach 1:

Instead of applying a single uniform downsampling factor across the entire image, the patent implements location-dependent downsampling ratios. The downsampling factor varies spatially, with central regions experiencing less downsampling and peripheral regions experiencing more downsampling. This maintains processing simplicity through automated algorithms while preserving image resolution in critical areas.

Inventive Principle:

Principle #3Local quality

3Manufacturing precision

If higher-level codecs are used to maintain image quality, then image resolution and quality are preserved, but costs increase

Engineering Contradiction:

Improveimage resolution and qualityVSAvoidcosts

Core Design Contradiction:

Manufacturing precisionVSEase of manufacture

Solution Approach 1:

The patent changes the approach from using higher-level codecs (which increase cost) to implementing non-uniform downsampling algorithms that work with existing standard codecs. By optimizing the input image through selective regional downsampling, the system achieves good image quality compatibility with widely deployed hardware and software infrastructure, thereby reducing costs while maintaining acceptable image resolution.

Inventive Principle:

Principle #35Parameter changes

4Adaptability or versatility

If more pixels are allocated to peripheral fields of view in EAC format, then uniform information density is assumed, but actual captured images from fisheye lenses do not have uniform information density

Engineering Contradiction:

Improveuniform information density assumptionVSAvoidnon-uniform information density mismatch

Core Design Contradiction:

Adaptability or versatilityVSLoss of information

Solution Approach 1:

Instead of assuming uniform information density and allocating pixels uniformly as in traditional EAC format, the patent inverts the approach by allocating pixels non-uniformly based on actual information density measurements. Regions with higher information density receive more pixels, while regions with lower information density receive fewer pixels, thereby reducing information loss and matching the actual characteristics of fisheye lens captures.

Inventive Principle:

Principle #13The other way round (Inversion)

Data Source

PatentUS10997693B2Apparatus and methods for non-uniform processing of image data

Publication Date: 2021.05.04 GOPRO INC

AI summary

Methods and apparatus for processing of high resolution content so as to obey desired encoder constraints. In one embodiment, the method includes capturing high resolution imaging spherical content; mapping the spherical content to another frame of reference (e.g., a non-uniform mapping and scaling) splitting up the mapped and scaled content into respective portions; feeding the split up portions to respective imaging encoders; packing encoded content from the respective imaging encoders into an A/V container; and storing and/or transmitting the A/V container. In one variant, the mapping and scaling are chosen to enable rendering of 1080P content in a desired scope or range (e.g., 360 degrees) using commodity encoder hardware and software.