Optical Flow Confidence Map for Low Light Conditions

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Optical flow estimation in advanced driver assistance systems (ADAS) faces challenges such as occlusions, low visibility, noise, and camera rotations, leading to inaccurate flow estimates, necessitating a method to generate a confidence map to assess the reliability of optical flow values.

Innovation Solution

A method that computes a confidence map by using a quasi-parametric approach, processing features like image and flow gradients, census texture gradients, and matching costs through a decision tree classifier to assign weights or ignore potentially incorrect optical flow values on a pixel-by-pixel basis.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If optical flow estimation is performed in low light or low texture regions, then motion detection coverage is improved, but measurement precision deteriorates due to incorrect flow estimates

Engineering Contradiction:

Improvemotion detection coverageVSAvoidflow estimation accuracy

Core Design Contradiction:

Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The patent applies local quality by computing census texture gradient specifically for low light or low texture regions where optical flow estimation is unreliable. This localized feature computation allows the system to identify problematic regions without affecting the entire image, enabling selective handling of low-quality flow estimates while maintaining overall system performance.

Inventive Principle:

Principle #3Local quality

2Reliability

If a confidence map is computed to assess flow estimation reliability, then reliability of higher-level algorithms is improved, but device complexity increases

Engineering Contradiction:

Improvealgorithm reliabilityVSAvoidcomputational complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent implements partial action by computing confidence assessment only where necessary - specifically in low light or low texture regions identified through census texture gradient. Rather than computing confidence maps for the entire image, the system selectively applies confidence assessment to regions where optical flow estimation is most likely to fail, reducing overall computational complexity while maintaining reliability where it matters most.

Inventive Principle:

Principle #16Partial or excessive action

3Productivity

If real-time optical flow estimation is performed, then productivity is improved, but measurement precision deteriorates due to challenging conditions like occlusions and noise

Engineering Contradiction:

Improvereal-time processing speedVSAvoidflow estimation accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent applies preliminary action by computing census texture gradient and identifying low light or low texture regions before performing optical flow estimation. This pre-assessment allows the system to prepare appropriate handling strategies for problematic regions, enabling real-time processing while maintaining precision through proactive identification of challenging areas.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS9811917B2Confidence estimation for optical flow under low light conditions

Publication Date: 2017.11.07 TEXAS INSTRUMENTS INC

AI summary

A confidence map for optical flow gradients is constructed calculating a set of census texture gradients for each pixel of an image, filtering said gradients and extracting confidence values from said gradients using a plurality of decision tree classifiers. A confidence map is then generated from said confidence values.