Image Sensor Vision Pixels Segmentation for Power Efficiency

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Digital devices face high power consumption and slow computation times due to excessive image pixel processing for vision recognition, which limits their application in portable devices and affects low-light performance.

Innovation Solution

An image sensor optimized for vision recognition, where specific 'vision pixels' are used for recognition tasks, reducing power consumption and enabling rapid processing, while all pixels are used for image capturing, with separate digital signal converters and processors for each function.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If all image pixels are processed for vision recognition, then recognition accuracy is improved, but power consumption increases substantially

Engineering Contradiction:
Improverecognition accuracyVSAvoidpower consumption
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The image sensor pixels are segmented into two functional groups: vision recognition pixels (first pixel array) and image capturing pixels (second pixel array). This segmentation allows the system to process only the necessary pixels for each function, reducing overall power consumption while maintaining recognition accuracy.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent extracts and separates the vision recognition function from the general image capturing function by dedicating specific pixels (first pixel array) to vision recognition tasks. This extraction enables independent optimization of each function's resource usage.

Inventive Principle:
Principle #2Taking out (Extraction)

2Loss of information

If all image pixels are processed for vision recognition, then recognition completeness is improved, but computation time increases

Engineering Contradiction:
Improverecognition completenessVSAvoidcomputation time
Core Design Contradiction:
Loss of informationVSLoss of time

Solution Approach 1:

By segmenting pixels into dedicated vision recognition pixels, the system processes only the necessary subset of pixels for recognition tasks, significantly reducing computation time while maintaining complete recognition information.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies partial action by processing only the specific first pixel array for vision recognition rather than all pixels, achieving sufficient recognition completeness with reduced computational burden.

Inventive Principle:
Principle #16Partial or excessive action

3Adaptability or versatility

If color filters are used in image sensor, then color image capturing is enabled, but low-light performance deteriorates

Engineering Contradiction:
Improvecolor image capturingVSAvoidlow-light performance
Core Design Contradiction:
Adaptability or versatilityVSReliability

Solution Approach 1:

Different pixel regions are assigned different optical characteristics: the first pixel array uses color filters for color vision recognition, while the second pixel array uses high-sensitivity white pixels (without color filters) for low-light image capturing, optimizing each region for its specific function.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent changes the optical parameters of different pixel regions by applying color filters to some pixels and leaving others as high-sensitivity white pixels, allowing the system to adapt to different lighting conditions and functional requirements.

Inventive Principle:
Principle #35Parameter changes

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

This approach allows for efficient vision recognition with low power consumption and fast computation, maintaining normal image capturing performance, and improves low-light sensitivity by using high-sensitivity white pixels.

Implementation Method 1

an image sensor... for converting optical energy into an electrical signal

Methodology Applied
Scientific EffectPhotoelectric Effect: Photoelectric Effect

Data Source

PatentEP2587407B1Vision recognition apparatus and method
Publication Date: 2020.03.04 SAMSUNG ELECTRONICS CO LTD
  • EP2587407B1 patent drawingFigure 1
  • EP2587407B1 patent drawingFigure 2
  • EP2587407B1 patent drawingFigure 3~4

AI summary

Some of pixels in a general image sensor for image capturing are designated as vision pixels used for vision recognition. Optical information obtained only from the vision pixels is used to perform vision recognition in the vision recognition mode. Capturing image data is generated based on optical information obtained from all the pixels in the image data in the image-shooting mode. A digital signal converter that converts optical information obtained only from vision pixels into a digital signal is separately provided in addition to a digital signal converter that converts optical information of all pixels in the image sensor into a digital signal.