Adaptive Disparity Estimation for Stereo Image Recognition

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing methods for estimating disparity in stereo images face challenges in determining optimal smooth energy coefficients, leading to noise in under-smooth cases and over-smoothing, which varies with image resolution.

Innovation Solution

A method that adapts smooth energy coefficients based on image resolution levels using a hierarchical graph cut algorithm, calculating coefficients through texture analysis and iterative refinement to produce accurate disparity maps.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If a fixed smooth energy coefficient is used in graph cut algorithm, then the algorithm is simple to implement, but the disparity map quality deteriorates due to over-smoothing or under-smoothing depending on image resolution

Engineering Contradiction:

Improvealgorithm complexityVSAvoiddisparity map quality

Core Design Contradiction:

Device complexityVSManufacturing precision

Solution Approach 1:

The patent applies dynamics by making the smooth energy coefficient adaptive rather than fixed. The coefficient dynamically adjusts based on image resolution and texture characteristics, transitioning from a static value to a dynamic parameter that optimizes disparity estimation quality across different imaging conditions.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The patent changes the parameter of smooth energy coefficient based on image resolution levels and texture analysis. By varying this parameter across different resolution levels and image regions, the system achieves optimal disparity map quality without requiring a completely different algorithm for each condition.

Inventive Principle:

Principle #35Parameter changes

2Stability of the object's composition

If high smooth energy coefficient is used to reduce noise, then disparity map smoothness improves, but object discontinuities are lost due to over-smoothing

Engineering Contradiction:

Improvedisparity map smoothnessVSAvoidobject boundary accuracy

Core Design Contradiction:

Stability of the object's compositionVSManufacturing precision

Solution Approach 1:

The patent applies local quality by assigning different smooth energy coefficients to different regions of the image based on texture analysis. Regions with high texture variability receive lower coefficients to preserve boundaries, while smooth regions receive higher coefficients for noise reduction, achieving optimal local characteristics throughout the image.

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The system dynamically adjusts the smooth energy coefficient based on local texture characteristics and resolution levels. This dynamic adaptation allows the system to reduce noise in smooth regions while preserving object discontinuities in textured regions, preventing over-smoothing artifacts.

Inventive Principle:

Principle #15Dynamics

3Manufacturing precision

If different smooth energy coefficients are calculated for different resolution images, then disparity map accuracy improves, but calculation time increases

Engineering Contradiction:

Improvedisparity map accuracyVSAvoidcalculation time

Core Design Contradiction:

Manufacturing precisionVSLoss of time

Solution Approach 1:

The patent applies preliminary action by pre-calculating texture characteristics and determining appropriate smooth energy coefficients for different resolution levels before executing the main disparity estimation. This preparation step enables efficient processing during the actual computation by avoiding real-time recalculations.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system dynamically selects smooth energy coefficients based on the resolution level and texture analysis performed during preprocessing. This dynamic selection strategy allows the system to use computationally expensive coefficient calculations only when necessary, reducing overall processing time while maintaining accuracy.

Inventive Principle:

Principle #15Dynamics

4Measurement precision

If graph cut algorithm is applied to high-resolution images directly, then disparity estimation detail is high, but computational complexity and time consumption increase significantly

Engineering Contradiction:

Improvedisparity estimation detailVSAvoidprocessing speed

Core Design Contradiction:

Measurement precisionVSProductivity

Solution Approach 1:

The patent applies segmentation by dividing the image processing into multiple resolution levels. The system processes images at different resolutions separately, using coarser resolution for initial disparity estimation and refining only in critical regions at higher resolutions, thereby reducing overall computational complexity.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system performs preliminary disparity estimation at lower resolution levels before proceeding to higher resolutions. This preliminary action provides a coarse disparity map that guides subsequent fine-tuning operations, reducing the computational burden at each stage while maintaining detailed accuracy where needed.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS7711181B2Method of estimating disparity for 3D object recognition

Publication Date: 2010.05.04 SAMSUNG ELECTRONICS CO LTD

AI summary

There is provided a method of estimating disparity for 3D object recognition. The method includes obtaining a plurality of images having different resolutions for the stereo image, estimating a disparity map for a lowest-resolution level image, estimating a coarse disparity map for an upper resolution level image by using the disparity, obtaining a fine disparity map for the upper resolution level image by using the coarse disparity, and outputting the fine disparity map as a final disparity map for the stereo image if the upper resolution level image has a resolution of a highest level.