Dual-Pixel Sub-Pixel Alignment via Bilinear Measures

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing dual-pixel sensor image processing techniques require significant computational resources for sub-pixel alignment, making them inefficient and not suitable for real-time applications.

Innovation Solution

The proposed method involves parametrically fitting signals from dual-pixel sensors, generating bilinear measures, and determining alignment confidence to estimate sub-pixel alignment in a single step, reducing computational complexity.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional sub-pixel alignment techniques are used, then measurement precision is improved, but computational resources and processing time increase significantly

Engineering Contradiction:

Improvesub-pixel alignment precisionVSAvoidprocessing speed

Core Design Contradiction:

Measurement precisionVSProductivity

Solution Approach 1:

The patent changes the mathematical parameters of the alignment process by using a simplified linear model instead of complex non-linear optimization. The alignment is achieved through a single matrix multiplication operation that computes the optimal shift directly from the dual-pixel signals, eliminating the need for iterative optimization and significantly reducing computational complexity while maintaining precision

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent replaces the mechanical iterative search process with a direct mathematical solution. Instead of iteratively adjusting alignment parameters and evaluating error metrics, the system uses a closed-form solution involving matrix operations to compute the alignment shift in a single step, substituting computational mechanics with algebraic manipulation

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If iterative optimization methods are used for sub-pixel alignment, then alignment accuracy is improved, but computation time and processing complexity increase

Engineering Contradiction:

Improvealignment accuracyVSAvoidprocessing complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

Instead of iteratively optimizing alignment by adjusting parameters and evaluating error metrics, the patent inverts the approach by directly computing the alignment shift through a single matrix operation. The solution is derived by inverting the relationship between the dual-pixel signals and the alignment parameter, allowing direct calculation without iteration

Inventive Principle:

Principle #13The other way round (Inversion)

Solution Approach 2:

The patent extracts the essential information needed for alignment directly from the dual-pixel signals through a single linear operation. By separating the alignment computation from the imaging process and extracting only the necessary correlation information, the system eliminates unnecessary computational steps while maintaining accuracy

Inventive Principle:

Principle #2Taking out (Extraction)

Data Source

PatentUS20250157062A1Apparatus and method for generating depth maps from raw dual pixel sensor data

Publication Date: 2025.05.15 CANON KK

AI summary

An image processing apparatus and method for generating a sub-pixel alignment estimate from dual-pixel sensor data is provided. The method includes parametrically fitting at least two signals observed across a two dual-pixels of the dual-pixel sensor in the direction of the dual-pixel split. generating at least one bilinear measure on the at least two signals' fitting parameters. determining an alignment confidence based at least in part on the at least one bilinear measure on the at least two signals' fitting parameters, and determining the sub-pixel alignment estimate based at least in part on the at least one bilinear measure on the at least two signals' fitting parameters.