Saliency Map Creation Using Human Visual System Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for creating saliency maps in images rely on non-psycho visual features, which do not effectively mimic the human visual system, limiting their ability to detect salient points efficiently.
Innovation Solution
A method that creates a temporal saliency map by decomposing images into frequential sub-bands, estimating movement, and combining this with spatial saliency maps using a model based on the human visual system, including steps like perceptual sub-band decomposition, masking, and weighting with the maximum pursuit velocity of the eye.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If non-psycho visual features are used for saliency map creation, then the computational process is simpler, but the detection accuracy of salient points deteriorates
Solution Approach 1:
The patent transforms the saliency detection approach by changing the parameter basis from non-psycho visual features to psycho-visual features that model human visual system characteristics. This includes using perceptual sub-band decomposition, masking functions, and motion estimation weighted by maximum pursuit velocity, thereby improving detection accuracy while maintaining computational feasibility through efficient algorithms.
2Measurement precision
If psycho-visual features based on human visual system are used, then the detection accuracy of salient points is improved, but the device complexity increases
Solution Approach 1:
The patent applies segmentation by decomposing the image into perceptual sub-bands based on human visual system characteristics. This breakdown allows complex psycho-visual processing to be divided into manageable stages: frequency decomposition, masking application, motion estimation, and saliency map generation, reducing overall processing complexity while maintaining high detection accuracy.
Solution Approach 2:
The patent performs preliminary actions by pre-decomposing images into perceptual sub-bands and pre-computing masking functions before saliency detection. Motion estimation is also performed in advance with weighting by maximum pursuit velocity, preparing processed data that simplifies the final saliency map generation step and reduces real-time processing complexity.
3Measurement precision
If hierarchical decomposition into frequential sub-bands is performed, then the temporal saliency detection is improved, but the processing time increases
Solution Approach 1:
The patent employs periodic action through its multi-resolution pyramid structure, where processing is organized into discrete levels (full resolution, half resolution, quarter resolution, eighth resolution). This hierarchical periodic processing allows temporal saliency detection to be performed systematically at different scales, improving detection precision while managing processing time through efficient multi-scale analysis.
Data Source
AI summary
Detection of the salient points in an image enable the improvement of further steps such as coding or image indexing, watermarking, video quality estimation. The methods rely on the fact that a model is fully based on the human visual system (HVS) such as the computation of early visual features, and the methods compute a saliency map for video images taking into account motion and the velocity of the eye.