Image Processing Device With Two-Stage Blur Correction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image processing devices face high processing loads and delays in real-time correction due to the need for extensive geometric calculations to correct intra-frame blur in high-frame-rate image stabilization, making it impractical to perform continuous shooting at higher frame rates.

Innovation Solution

The device employs a two-stage correction process: first, translation processing is applied to all captured images based on gyro information, followed by addition processing to generate an addition image, and then higher-load distortion and rotation corrections are applied only to the addition image, reducing overall processing time and load.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If continuous shooting is performed at a high frame rate and correction processing is performed on each captured image, then intra-frame blur can be reduced, but the processing load increases and real-time processing becomes difficult

Engineering Contradiction:
Improveblur reductionVSAvoidprocessing speed
Core Design Contradiction:
Manufacturing precisionVSProductivity

Solution Approach 1:

The correction processing is divided into two distinct stages: first correction processing performed on each captured image at high frame rate, and second correction processing performed on the addition image at lower frame rate. This segmentation allows the system to maintain high processing speed while achieving comprehensive blur correction.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The first correction processing is performed preliminarily on each captured image to generate corrected images, which are then used in the addition processing. This preliminary action prepares the data for the second correction processing stage, enabling efficient real-time processing.

Inventive Principle:
Principle #10Preliminary action

2Manufacturing precision

If correction processing is performed on each of N captured images, then image stabilization quality improves, but processing time increases

Engineering Contradiction:
Improveimage stabilization qualityVSAvoidprocessing time
Core Design Contradiction:
Manufacturing precisionVSLoss of time

Solution Approach 1:

The addition processing merges N corrected images into a single addition image, reducing the number of images that require second correction processing from N to 1. This merging significantly reduces processing time while maintaining stabilization quality.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The addition image serves multiple functions: it combines the results of first correction processing, reduces the burden on the second correction processing, and provides a basis for final output. This multi-functionality optimizes the overall processing efficiency.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentEP3905656B1Image processing device
Publication Date: 2025.09.24 SONY SEMICON SOLUTIONS CORP
  • EP3905656B1 patent drawingFigure 1(A)~1(B)
  • EP3905656B1 patent drawingFigure 2
  • EP3905656B1 patent drawingFigure 3(A)~3(C)

AI summary

An image processing device of the present disclosure includes: a first correction processing section that performs, on each of a plurality of captured images captured by an imaging device, first correction processing on the basis of posture information of the imaging device and generates a plurality of first corrected images; an addition processing section that performs addition processing on at least two first corrected images among the plurality of first corrected images generated by the first correction processing section and generates an addition image; and a second correction processing section that performs, on the addition image generated by the addition processing section, second correction processing higher in processing load than the first correction processing and generates a second corrected image.