Image Processing for Multi-Part Subject Focus Selection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing imaging technologies face challenges in accurately detecting and prioritizing the main subject for focus adjustment, particularly when the characteristic parts of a subject, such as the whole body or trunk, are difficult to detect due to changes in posture or clothing, leading to unreliable focus positioning.

Innovation Solution

An image processing device with detection units for identifying multiple subject parts (head, eye, and trunk) and an association unit to determine a target subject, ensuring that if the initially detected part is not found, the device switches to another part for focus, thereby improving focus accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If multiple characteristic parts of a subject are detected to increase focusing success rate, then the reliability of detection improves, but the accuracy of determining the main subject for focus decreases

Engineering Contradiction:
Improvedetection reliabilityVSAvoidmain subject determination accuracy
Core Design Contradiction:
ReliabilityVSMeasurement precision

Solution Approach 1:

The patent applies local quality by differentiating the treatment of different body parts based on their importance for main subject determination. Head and eye regions are prioritized over trunk regions, with specific logic pathways that favor head/eye detections. This selective prioritization resolves the contradiction by ensuring that while multiple parts are detected (improving reliability), the determination logic specifically weights certain parts heavier to maintain accuracy in identifying the main subject.

Inventive Principle:
Principle #3Local quality

2Adaptability or versatility

If detection uses low reliability threshold to detect at least one part, then detection coverage improves, but focus accuracy on the correct subject decreases

Engineering Contradiction:
Improvedetection coverageVSAvoidfocus accuracy
Core Design Contradiction:
Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The patent implements dynamics through a multi-stage detection and determination process. First, detection operates with a low threshold to maximize coverage. Then, the determination unit dynamically adjusts its logic based on what was detected - if head/eye parts are detected, they are prioritized; if only trunk is detected, alternative logic applies. This dynamic adaptation allows the system to maintain both broad detection coverage and accurate focus selection.

Inventive Principle:
Principle #15Dynamics

3Reliability

If the system switches to another part for focus when the first part is not detected, then focus reliability improves, but the complexity of determination logic increases

Engineering Contradiction:
Improvefocus reliabilityVSAvoiddetermination logic complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent applies segmentation by dividing the determination logic into distinct pathways based on detected body parts. Separate determination flows are established for head/eye detections versus trunk detections. This modular segmentation of the determination logic reduces overall complexity by making each specific case straightforward, while the high-level structure maintains reliability through the backup mechanism of switching to alternative parts when primary detections fail.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentUS20250336232A1Image processing device, control method thereof, imaging device, and program storage medium that determine a region of an image including detected first and second regions to be a target subject region
Publication Date: 2025.10.30 CANON KK
  • US20250336232A1 patent drawing
  • US20250336232A1 patent drawing
  • US20250336232A1 patent drawing

AI summary

An image processing device that determines a target subject region to be processed from an image for a specific subject. The device includes at least one processor and at least one memory functioning as a first detection unit configured to detect a first region corresponding to a part having a first feature of the specific subject from the image, a second detection unit configured to detect a second region corresponding to a part having a second feature of the specific subject from the image, an association unit configured to associate the first region detected by the first detection unit and the second region detected by the second detection unit, and a determination unit configured to determine any region of the image including the first region detected by the first detection unit and the second region detected by the second detection unit as the target subject region.