Image Composition Evaluation via Region Segmentation and Attribution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image composition evaluation methods, such as Fujifilm's Human Composition Judgement, are limited in evaluating images without human faces, landscape images, and other types of composition problems, as they focus primarily on human object placement and do not consider background or scale, leading to incomplete composition assessments.
Innovation Solution
An image composition evaluating apparatus and method that segment images into regions, extract attributes from each region, describe relationships among them, and evaluate composition based on these attributes and preset criteria, enabling the evaluation of various image types and composition issues like object placement, area size, and color matching.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If human face detection and gravity center evaluation methods are used, then human object placement can be evaluated, but images without human faces (landscape images, object pictures, human backside pictures) cannot be evaluated at all
Solution Approach 1:
The patent replaces the specialized human-face-detection-based evaluation method with a universal region-based evaluation method that can handle all types of images including landscapes, object pictures, and images without human faces. The region segmentation and attribution extraction process works uniformly across different image types, making the evaluation system multi-functional and adaptable to diverse image content.
Solution Approach 2:
The patent segments the image into multiple regions and evaluates each region's attribution independently, then integrates these evaluations to assess overall composition. This segmentation approach allows the system to handle images without relying on human face detection, as each segmented region can be evaluated based on its own characteristics and relationships with other regions.
2Measurement precision
If only human object placement is considered, then center region positioning can be achieved, but other composition problems (object scale, background proportion, object relationships) cannot be evaluated
Solution Approach 1:
The patent extends the evaluation from a single dimension (human object placement) to multiple dimensions by introducing region attribution extraction that considers scale, position, color, texture, and other characteristics. This multi-dimensional evaluation framework enables comprehensive assessment of various composition problems including object scale, background proportion, and relationships between different image elements.
Solution Approach 2:
By segmenting the image into multiple regions and evaluating each region's attribution separately, the system can assess different composition aspects (scale, position, color matching, etc.) for each region and their relationships, thereby achieving comprehensive evaluation of multiple composition problems simultaneously.
3Measurement precision
If region segmentation and attribution extraction are performed for all regions, then comprehensive evaluation of diverse image types is achieved, but processing complexity and computational load increase
Solution Approach 1:
The patent extracts only the essential attributions (scale, position, color, texture, etc.) from each segmented region that are necessary for composition evaluation, rather than analyzing all possible image characteristics. This selective extraction of key attributes maintains evaluation accuracy while reducing processing complexity compared to comprehensive image analysis.
Data Source
AI summary
In the present invention, an attribution is extracted from each region obtained by segmentation of an image, relationships among the regions are described, and a composition of the image is evaluated based on the attributions and the relationships.