Main Subject Detection via Multi-Saliency Feature Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting the main subject in images are not completely satisfactory, lacking in accuracy and precision, which affects applications such as auto-focus, auto-exposure, image recognition, and image editing.
Innovation Solution
A method that computes multiple saliency features like sharp/blur, spectral residual, color spatial distribution, and color contrast features, generates a spatial weight map, adjusts and combines these features to create a saliency map, and refines the detection through multiple stages to accurately identify the main subject.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing methods for main subject detection are used, then the detection process is simple, but the accuracy and precision of main subject detection are insufficient
Solution Approach 1:
The image is divided into multiple image segments, and saliency features are computed for each segment separately. This segmentation approach allows the algorithm to process complex images in manageable parts, improving detection accuracy while maintaining computational feasibility through localized analysis of sharp/blur, spectral residual, color spatial distribution, and color contrast features in each segment.
Solution Approach 2:
The patent employs multiple saliency features (sharp/blur, spectral residual, color spatial distribution, color contrast) that capture different parameter aspects of the image. By computing and combining these diverse parameters, the algorithm achieves higher detection precision through multi-parameter analysis rather than relying on a single feature type.
2Measurement precision
If multiple saliency features are computed and combined, then the main subject detection accuracy improves, but the computational complexity increases
Solution Approach 1:
By segmenting the image first and computing saliency features for each segment independently, the algorithm reduces the computational burden compared to processing the entire image at once. This segmentation strategy enables parallel processing of multiple features across segments, improving processing efficiency while maintaining the benefits of multi-feature analysis for accurate detection.
Solution Approach 2:
The patent computes multiple saliency features (sharp/blur, spectral residual, color spatial distribution, color contrast) for each segment, which may seem excessive, but this partial computation approach allows selective combination of the most discriminative features. This strategy achieves high detection accuracy by focusing computational resources on the most informative features for each image segment.
Data Source
AI summary
A method for detecting a main subject in an image comprises the steps of: (i) computing a plurality of saliency features from the image (14) with a control system (20); (ii) generating a spatial weight map for each of a plurality of image segments of the image (14) with the control system (20); (iii) adjusting the plurality of saliency features via the spatial weight map to generate a plurality of adjusted saliency features; (iv) combining at least two of the plurality of adjusted saliency features to generate a saliency map of the image (14); and (v) extracting the main subject from the saliency map of the image (14). Additionally, the plurality of saliency features can include at least two of a sharp/blur saliency feature, a spectral residual saliency feature, a color spatial distribution saliency feature, and one or more color contrast saliency features.