Image Saliency Detection via Regional Contrast Segmentation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional image processing methods are inefficient in rapidly and uniformly highlighting salient object regions in images, particularly for real-time applications and complex natural images.

Innovation Solution

An image processing method and device that calculates visual saliency based on regional contrast by segmenting images into regions and computing a weighted sum of color differences between regions, using a formula that incorporates spatial and color distance factors.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If local contrast based methods are used to compute saliency by rarity of image regions in small local neighborhoods, then edge regions are highlighted with higher saliency values, but the entire visual-salient objects are not uniformly highlighted

Engineering Contradiction:
Improvesaliency detection precisionVSAvoiduniformity of saliency distribution
Core Design Contradiction:
Measurement precisionVSStability of the object's composition

Solution Approach 1:

The image is segmented into multiple superpixels or regions using algorithms like SLIC (Simple Linear Iterative Clustering). This segmentation groups adjacent pixels into meaningful regions, allowing the method to evaluate saliency at the region level rather than pixel level, which promotes uniform highlighting while maintaining edge detection capability

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent transitions from local neighborhood comparison to global region-based comparison by introducing a new dimension of evaluation. Instead of comparing each pixel only with its local neighbors, the method compares segmented regions across the entire image using color histogram differences, thereby achieving both edge detection and uniform object highlighting

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Productivity

If global contrast based methods are used to evaluate saliency by measuring difference between image region and entire image, then computational efficiency is improved, but distinctiveness clues in other color channels are ignored

Engineering Contradiction:
Improveprocessing speedVSAvoidcolor channel information
Core Design Contradiction:
ProductivityVSLoss of information

Solution Approach 1:

The patent employs multiple color channels (RGB or LAB color spaces) simultaneously to evaluate regional contrast. By computing color histogram differences across all color channels, the method captures distinctiveness clues from multiple color dimensions, making the saliency detection more comprehensive while maintaining global contrast efficiency

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The saliency computation uses a composite approach by combining color histogram information from multiple color channels. The color difference metric integrates information from different color spaces, creating a composite representation that captures more comprehensive color distinctiveness while maintaining computational efficiency

Inventive Principle:
Principle #40Composite materials

3Measurement precision

If learning-based automatic detection methods are used for salient objects in videos, then detection accuracy is improved, but processing time increases to several seconds

Engineering Contradiction:
Improvedetection accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent performs preliminary image segmentation into superpixels or regions before saliency computation. This preprocessing step organizes the image data into meaningful units, which accelerates the subsequent saliency calculation by reducing the number of comparisons needed while maintaining detection accuracy

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent replaces complex learning-based mechanical systems with a simpler physics-inspired approach based on color histogram comparison and regional contrast. This substitution maintains detection accuracy while dramatically reducing processing time, enabling real-time application

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentEP2523165B1Image processing method and image processing device
Publication Date: 2017.09.27 OMRON CORP
  • EP2523165B1 patent drawingFigure 1
  • EP2523165B1 patent drawingFigure 2
  • EP2523165B1 patent drawingFigure 3

AI summary

The present invention relates to an image processing method and image processing device for detecting visual saliency of an image based on regional contrast. The method includes: a segmentation step that segments an input image into a plurality of regions by using an automatic segmentation algorithm; and a computation step that calculates a saliency value of one region of the plurality of segmented regions by using a weighted sum of color differences between the one region and all other regions. According to the present invention, it is possible to automatically analyze visual saliency regions in an image, and a result of analysis can be used in application areas including significant object segmentation, object recognition, adaptive image compression, content-aware image resizing, and image retrieval.