Providing an image of an object from a refrigeration appliance

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The limited optical capabilities of cameras in refrigerators result in low-quality images, which are often blurry or noisy, and existing methods like AI enhancements cannot improve image quality beyond the original recording, hindering object detection and image processing.

Innovation Solution

A method that captures multiple images from comparable perspectives and combines them to improve sharpness and reduce noise, allowing for a larger camera-object distance and using sub-pixel alignment and machine learning for image recognition, while maintaining a consistent exposure to enhance image quality.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Area of stationary object

If multiple images are captured from different perspectives to create a panoramic view, then the overall view coverage is improved, but the image sharpness and quality deteriorate due to perspective differences

Engineering Contradiction:
Improveview coverageVSAvoidimage sharpness
Core Design Contradiction:
Area of stationary objectVSMeasurement precision

Solution Approach 1:

The patent divides the image improvement task into two separate processes: first capturing multiple images from different perspectives for panoramic coverage, then performing post-processing alignment and combination to achieve sharpness. This segmentation allows each process to optimize for its specific goal without compromise.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary actions of capturing multiple images with sufficient overlap before final combination. By pre-capturing images with consistent exposure settings and sufficient temporal overlap, the system prepares the necessary data foundation for subsequent sharp image combination, resolving the contradiction between coverage and sharpness.

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If AI methods like GAN or Variational Auto-Encoders are used to increase resolution, then the image resolution is improved, but the information content cannot exceed that of the original recording

Engineering Contradiction:
Improveimage resolutionVSAvoidinformation content
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The patent combines multiple actual image recordings that capture the same object from comparable perspectives. By merging these real recordings, the system achieves higher effective resolution and reduced noise while preserving genuine information content from the original scenes, avoiding the information loss inherent in AI-generated enhancements.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent uses multiple copies (recordings) of the same scene taken at different times to reconstruct a higher quality image. Instead of generating information through AI, the system captures multiple real copies and combines them, ensuring that the information content remains authentic and complete.

Inventive Principle:
Principle #26Copying

3Measurement precision

If the camera is placed close to the object for detailed imaging, then the image detail is improved, but the field of view and overview capability deteriorate

Engineering Contradiction:
Improveimage detailVSAvoidfield of view
Core Design Contradiction:
Measurement precisionVSArea of stationary object

Solution Approach 1:

The patent transitions from spatial proximity (placing camera close to object) to temporal proximity (capturing multiple images at different times). By moving the dimension of image acquisition from space to time, the system can maintain a fixed camera position with sufficient field of view while still achieving detailed imaging through multiple captures and combination.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

Solution Approach 2:

The patent performs preliminary capture of multiple images that include the entire object within the field of view, then processes these images to extract and combine the detailed portions. This preliminary action ensures comprehensive coverage is maintained while detailed information is recovered through post-processing.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4279847A1Providing an image of an object from a refrigeration appliance
Publication Date: 2023.11.22 BSH HAUSGERATE GMBH
  • EP4279847A1 patent drawingFigure 1
  • EP4279847A1 patent drawingFigure 2
  • EP4279847A1 patent drawingFigure 3

AI summary

A method (200) for providing an image of an object (135) in a cooling device (100) comprises steps of capturing a first image (305) of the object (135) using a camera (130) from a predetermined perspective; determining a second image (310) that was captured from a comparable perspective at an earlier time of the object (135); and combining (230) the first and second images (305, 310) to improve image sharpness and/or reduce image noise in a resulting combined image.