Dual-Camera Feature Extraction for Low-Complexity Image Processing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing feature extraction methods from images and video frames are computationally intensive and time-consuming, limiting the speed and accuracy of analysis.

Innovation Solution

A dual-sensor setup combining an event-driven camera and a conventional CMOS camera is used to extract features, where motion statistics and intensity change information from the event-driven camera are utilized to identify objects of interest, allowing for efficient feature extraction directly from the CMOS camera output without additional computations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If conventional feature extraction methods are used on standard image sensor output, then comprehensive image analysis is achieved, but computational complexity and processing time increase significantly

Engineering Contradiction:
Improvefeature extraction speedVSAvoidcomputational complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent extracts and processes only the most relevant features from event-driven camera data - specifically motion statistics and intensity change information - rather than processing all image data. This selective extraction of critical information reduces computational complexity while maintaining feature extraction speed and accuracy.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent segments the image processing task by using the event-driven camera to identify objects of interest first, then applying feature extraction only to those specific regions. This divides the overall processing into two stages: object identification via motion detection, followed by targeted feature extraction, thereby reducing overall computational load.

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If feature extraction is performed on the whole image, then all potential objects are covered, but processing time and computational resources increase

Engineering Contradiction:
Improvefeature extraction accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies different processing quality levels to different regions of the image. Objects identified by the event-driven camera receive full feature extraction processing, while other regions receive minimal or no processing. This local differentiation maintains accuracy for important objects while reducing overall processing time.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The event-driven camera performs preliminary identification of objects of interest before the main feature extraction process. This preliminary action filters the image data to identify only those regions requiring detailed analysis, allowing the subsequent feature extraction to focus computational resources on relevant areas and reduce overall processing time.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3827582B1Image processing apparatus and method for feature extraction
Publication Date: 2026.02.25 HUAWEI TECH CO LTD
  • EP3827582B1 patent drawingFigure 1a
  • EP3827582B1 patent drawingFigure 1b
  • EP3827582B1 patent drawingFigure 1c

AI summary

The invention relates to an image processing apparatus for extracting a feature from an image of a scene. The apparatus may comprises processing circuitry configured, in response to a feature extraction event, to extract a feature from first image data representing a first image of a scene, wherein the feature extraction event is based on second image data representing a second image of the scene. The image processing apparatus may include an event driven sensor and the second image may include a positive/negative amount of the intensity change and the location of the intensity change. Moreover, the invention relates to a corresponding image processing method.