Camera Autofocus ROI Filtering Before Demosaicing

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing autofocus techniques in camera devices are computationally expensive and lack accuracy due to intermixing of color channel characteristics during demosaicing, which affects the reliability of autofocus statistics.

Innovation Solution

Apply separate horizontal and vertical infinite impulse response (IIR) filters to image data before demosaicing, generating accumulation matrices to determine autofocus statistics in constant time using parallel processing, thereby reducing computational complexity and intermixing of color channel characteristics.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional autofocus filtering is applied to the entire camera image, then comprehensive focus information is obtained, but computational complexity increases and processing time extends

Engineering Contradiction:

Improveautofocus statistics accuracyVSAvoidcomputational complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent divides the camera image into multiple regions of interest (ROIs) and applies filtering operations only to these specific regions rather than the entire image. This segmentation approach maintains comprehensive focus information while significantly reducing computational complexity by limiting processing to only the most relevant areas containing potential focus targets.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies different processing strategies to different regions of the image based on their importance for autofocus determination. High-priority regions receive full filtering processing, while lower-priority regions use simplified processing or are excluded, optimizing the balance between measurement precision and computational complexity.

Inventive Principle:

Principle #3Local quality

2Measurement precision

If infinite impulse response (IIR) filters are applied to the camera image, then sharpness values are generated, but processing time increases and parallelization becomes difficult

Engineering Contradiction:

Improvesharpness value accuracyVSAvoidprocessing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent segments the image processing into independent region operations, allowing each region's IIR filtering to be performed separately. This enables parallel execution of filtering operations across multiple regions using multi-core processors or GPU architectures, significantly reducing total processing time while maintaining sharpness value accuracy.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary identification of regions of interest before applying computationally intensive IIR filtering. By pre-defining which regions require full processing based on simple criteria (such as detecting edges, corners, or high-contrast areas), the system prepares the processing pipeline in advance, allowing optimized parallel execution and reducing overall processing time.

Inventive Principle:

Principle #10Preliminary action

3Ease of operation

If demosaicing is performed before autofocus statistics determination, then full resolution color planes are available, but color channel intermixing reduces autofocus accuracy

Engineering Contradiction:

Improveimage processing convenienceVSAvoidautofocus statistics accuracy

Core Design Contradiction:

Ease of operationVSMeasurement precision

Solution Approach 1:

The patent inverts the conventional processing order by determining autofocus statistics from the raw Bayer format image data before performing demosaicing. This reversal prevents color channel intermixing that would occur during demosaicing, thereby maintaining higher autofocus statistics accuracy. The system processes the image in the order: raw Bayer data → autofocus statistics calculation → demosaicing, rather than the traditional demosaicing first approach.

Inventive Principle:

Principle #13The other way round (Inversion)

4Productivity

If region of interest processing is implemented, then processing speed increases, but comprehensive focus coverage may be reduced

Engineering Contradiction:

Improveprocessing speedVSAvoidfocus determination reliability

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The patent implements a multi-functional region processing system that can dynamically adapt to different autofocus scenarios. The same ROI processing framework handles both quick preview autofocus (using fewer, smaller regions) and precise capture autofocus (using more, larger regions). The system universally applies filtering and sharpness calculation algorithms across multiple regions, ensuring comprehensive focus coverage while maintaining high processing speed through efficient parallel execution.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS20260075326A1Regions of interest in autofocus processing

Publication Date: 2026.03.12 QUALCOMM INC

AI summary

One or more processors of a computing device may receive image data obtained via an image sensor of a camera. The one or more processors may apply one or more horizontal filters to the image data to generate horizontally filtered image data and may apply one or more vertical filters to the image data to generate vertically filtered image data. The one or more processors may generate a first accumulation matrix for the horizontally filtered image data and may generate a second accumulation matrix for the vertically filtered image data. The one or more processors may determine autofocus statistics based on the first accumulation matrix and the second accumulation matrix and may output information to adjust one or more lenses of the camera based on the autofocus statistics.