Face Feature Extraction for Accurate Illustration Generation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing face image recognition techniques require user operation to create similar face pictures, which may not fully capture the features of the original face image, leading to incomplete perception of the face features in the transformed image.

Innovation Solution

An information processing apparatus and method that acquires a target image, extracts a face region, identifies model face parts by comparing them to stored models, and determines an illustration image corresponding to the identified parts, allowing for the generation of a transformed image that perceives the features of the face image without user operation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If geometric sizes or ratios of part images are used to select transformed part images, then the selection process is simple, but the features of the face image are not completely perceived

Engineering Contradiction:
Improvefeature perception accuracyVSAvoidimage processing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by pre-storing multiple types of transformed part images (front view, profile view, three-quarter view) in the storage unit before the actual face image processing. This allows the system to quickly retrieve and select appropriate transformed images based on the detected face orientation, without needing to perform complex real-time transformations. The preliminary preparation of various view images enables accurate feature perception while maintaining processing efficiency.

Inventive Principle:
Principle #10Preliminary action

2Productivity

If user operation is required to create similar face picture images, then customization is possible, but the process becomes time-consuming and automated feature perception is lost

Engineering Contradiction:
Improveimage generation speedVSAvoiduser operation requirement
Core Design Contradiction:
ProductivityVSEase of operation

Solution Approach 1:

The system implements self-service by automatically detecting face part orientations (eyes, eyebrows, nose, mouth) and selecting appropriate transformed part images without requiring user intervention. The image generation apparatus autonomously completes the entire process from face image input to similar face picture output, enabling rapid automated feature perception and image generation while maintaining ease of use through simple face image input.

Inventive Principle:
Principle #25Self-service

3Measurement precision

If multiple transformed part images are stored for different views, then feature perception accuracy improves, but storage requirements and processing complexity increase

Engineering Contradiction:
Improveface feature accuracyVSAvoidimage data volume
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent applies segmentation by dividing the face image into distinct parts (eyes, eyebrows, nose, mouth) and storing transformed versions of each part separately for different views. This segmented approach allows the system to selectively process and recombine only the necessary face parts based on the detected orientation, rather than storing and processing entire face images for every possible view. This reduces overall data volume while maintaining high feature perception accuracy for each facial component.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP2315158B1Information processing apparatus, information processing method, and program
Publication Date: 2018.08.01 SONY GROUP CORP
  • EP2315158B1 patent drawingFigure 1
  • EP2315158B1 patent drawingFigure 2
  • EP2315158B1 patent drawingFigure 3A~3H

AI summary

An information processing apparatus including an image acquisition unit that acquires a target image; a face part extraction unit that extracts a face region including a face part from the target image; an identification unit that identifies a model face part by comparing the face part to a plurality of model face parts stored in a storage unit; and an illustration image determination unit that determines an illustration image corresponding to the identified model face part.