Saliency Map Computation via Boundary Region Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional techniques for computing saliency maps fail to retain high-frequency details of a scene and are computationally inefficient, limiting their suitability for various applications.
Innovation Solution
The method generates a base saliency map by combining intermediate saliency maps computed for boundary regions of an image, suppressing the effect of foreground objects that occupy a majority of these regions, and using these maps to derive additional saliency maps for post-processing actions such as image cropping suggestions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional techniques are used to compute saliency maps, then computational resources and time are consumed, but the resulting saliency maps are blurry and do not retain high-frequency details
Solution Approach 1:
The image is divided into multiple overlapping patches, and saliency maps are computed independently for each patch. This segmentation approach allows high-frequency details to be preserved in individual patches while the overall computational load is distributed and managed efficiently.
Solution Approach 2:
The patent transforms the traditional full-image saliency computation into a multi-scale patch-based computation. By working at different scales and combining results through a pyramid structure, the method achieves both high detail retention and computational efficiency by processing smaller units rather than the entire image at once.
2Reliability
If conventional techniques are used to compute saliency maps, then the computation can be performed, but the results are blurry and not suitable for applications requiring high-frequency details
Solution Approach 1:
The patent performs preliminary saliency computations on individual patches before combining them into the final saliency map. This preliminary processing at the patch level ensures that high-frequency details are captured early in the computation pipeline, making the final result suitable for applications requiring fine details.
Solution Approach 2:
Individual patch saliency maps are merged together using a pyramid-based combination strategy. This merging process integrates the high-frequency details from each patch while maintaining overall image coherence, producing a reliable saliency map suitable for various applications.
Data Source
AI summary
Saliency map computation is described. In one or more implementations, a base saliency map is generated for an image of a scene. The base saliency map may be generated from intermediate saliency maps computed for boundary regions of the image. Each of the intermediate saliency maps may represent visual saliency of portions of the scene that are captured in the corresponding boundary region. The boundary regions may include, for instance, a top boundary region, a bottom boundary region, a left boundary region, and a right boundary region. Further, the intermediate saliency maps may be combined in such a way that an effect of a foreground object on the saliency map is suppressed. The foreground objects for which the effect is suppressed are those that occupy a majority of one of the boundary regions.