Salient Object Detection via Composition Cost Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods, particularly those using saliency maps, are ineffective for locating salient objects in images with complex objects or backgrounds, often resulting in loss of image information.
Innovation Solution
A computing device determines saliency measures for image locations by calculating composition costs based on appearance and spatial distances of parts within a location from parts outside that location, using a sliding window to segment the image and identify local maxima as salient objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Difficulty of detecting and measuring
If saliency maps are used to locate salient objects, then the image can be processed to identify distinct parts, but image information is lost and detection effectiveness decreases for complex objects or backgrounds
Solution Approach 1:
The image is segmented into multiple parts using a sliding window approach, where each window position defines a location that is divided into inside parts (within the window) and outside parts (outside the window). This segmentation enables the system to analyze local regions while preserving global context information, avoiding the information loss associated with traditional saliency map transformations.
Solution Approach 2:
The patent introduces a new dimensional approach by calculating composition costs that consider both appearance distance and spatial distance between parts. Instead of transforming the image into a saliency map (2D transformation), the system operates in a multi-dimensional feature space that captures both visual similarity and spatial relationships, thereby preserving information while improving detection effectiveness.
2Adaptability or versatility
If traditional saliency map transformation is applied, then salient objects can be identified in simple images, but the method becomes ineffective for images with complex objects or backgrounds
Solution Approach 1:
The patent applies local quality by calculating composition costs specific to each location's inside parts based on their appearance and spatial distances from outside parts. This localized analysis allows the system to adapt to complex images by evaluating each region's unique characteristics rather than applying a uniform transformation, thereby maintaining effectiveness across diverse image types.
Solution Approach 2:
The system changes parameters by using composition costs derived from appearance distance and spatial distance metrics. Instead of relying on fixed saliency map thresholds, the method dynamically adjusts its analysis based on the specific visual and spatial characteristics of each location, enabling effective detection in complex images with varying object and background properties.
3Measurement precision
If composition cost is calculated for all image locations, then salient objects can be accurately identified, but computational complexity increases
Solution Approach 1:
The patent applies partial action by calculating composition costs only for locations where salient objects are likely present, rather than uniformly processing the entire image. The sliding window approach naturally limits the analysis to relevant regions, reducing computational complexity while maintaining accuracy in identifying salient objects.
Data Source
AI summary
A computing device configured to determine, for each of a plurality of locations in an image, a saliency measure based at least on a cost of composing parts of the image in the location from parts of the image outside of the location is described herein. The computing device is further configured to select one or more of the locations as representing salient objects of the image based at least on the saliency measures.