Image Scaling With Depth-Aware Region Segmentation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional image scaling algorithms do not account for the relative positions of objects during image formation, resulting in scaled images that do not accurately represent depth relations in the scene.

Innovation Solution

A method and apparatus that obtain depth values for both target and non-target regions in an image, calculate unique scaling ratios for each region based on their depth values, and apply these ratios to scale the image, ensuring that the scaled image maintains depth relations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If a same scaling ratio is used for scaling the image, then the scaling process is simple, but the scaled image does not accord with the depth relations in the scene

Engineering Contradiction:
Improvescaling process complexityVSAvoiddepth relation accuracy
Core Design Contradiction:
Device complexityVSManufacturing precision

Solution Approach 1:

The image is divided into multiple regions (target region and non-target region) based on depth values. Each region is assigned a different scaling ratio according to its depth, allowing differential scaling that preserves depth relationships while maintaining manageable processing complexity

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Different scaling ratios are applied to different regions of the image based on their depth values. The target region receives one scaling ratio while non-target regions receive different scaling ratios, creating local quality variations that accurately represent the three-dimensional depth structure of the scene

Inventive Principle:
Principle #3Local quality

Data Source

PatentEP2383701B1Image processing method and apparatus
Publication Date: 2014.05.21 HUAWEI DEVICE CO LTD
  • EP2383701B1 patent drawingFigure 1
  • EP2383701B1 patent drawingFigure 2
  • EP2383701B1 patent drawingFigure 3

AI summary

A method and an apparatus for processing an image are disclosed. The method includes: obtaining depth values of an image including a target region and a non-target region; obtaining a scaling ratio of the target region; obtaining a scaling ratio of the non-target region according to the depth values of the image and the scaling ratio of the target region; scaling respectively the target region and the non-target region according to the scaling ratio of the target region and the scaling ratio of the non-target region, and obtaining a scaled image. By implementing the present invention, a scaled image is an image according with depth relations in a scene can be realized.