Flash Detection in CMOS Image Sensors Using Line Average Luminance

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image processing technologies face challenges in accurately detecting and correcting discontinuous image brightness caused by strong, short-duration light sources like flashes in CMOS sensor images without increasing memory resources or incurring significant delays.

Innovation Solution

An image processing device and method that calculates line and screen average luminance, compares them to detect flash-induced discontinuities, and corrects image data by replacing affected lines with corresponding lines from a past screen, all while minimizing memory usage and processing delay.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If frame difference calculation is performed using line average luminance levels to detect external flash, then flash detection capability is improved, but false positives increase and processing complexity increases

Engineering Contradiction:
Improveflash detection accuracyVSAvoidprocessing complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent divides the screen into multiple line regions and calculates line average luminance levels for each line separately. This segmentation allows the system to identify localized flash effects (affecting only specific lines) versus global brightness changes (affecting the entire screen), thereby improving detection accuracy while maintaining manageable processing complexity through localized analysis.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Instead of performing full frame difference calculations across the entire image, the patent applies a simplified approach by comparing only line average luminance levels. This partial action reduces computational complexity while still effectively detecting flash events, as flashes typically affect only portions of the screen rather than the entire frame.

Inventive Principle:
Principle #16Partial or excessive action

2Measurement precision

If line average luminance calculation and frame difference calculation are performed for each line, then flash detection precision is improved, but processing time increases

Engineering Contradiction:
Improveflash detection precisionVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent segments the image processing into line-by-line luminance calculation, allowing precise detection of flash effects in specific regions while avoiding the need to process entire frames. This segmentation maintains high detection precision for localized flash events while reducing overall processing time compared to full-frame analysis.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent changes the parameter being analyzed from full-frame luminance to line-average luminance. This parameter transformation enables the system to detect flash events with high precision by focusing on the specific metric (line average) that directly reflects flash impact, while the simplified calculation reduces processing time requirements.

Inventive Principle:
Principle #35Parameter changes

3Reliability

If continuous frame comparison is performed to detect flash, then detection accuracy is improved, but memory resource requirements increase

Engineering Contradiction:
Improvedetection accuracyVSAvoidmemory resource
Core Design Contradiction:
ReliabilityVSQuantity of substance

Solution Approach 1:

The patent extracts only the essential information needed for flash detection by calculating and comparing line average luminance levels, rather than storing and processing complete frame data. This extraction approach maintains detection accuracy by preserving the key metric (luminance variation) while significantly reducing memory resource requirements by discarding redundant pixel-level information.

Inventive Principle:
Principle #2Taking out (Extraction)

4Manufacturing precision

If correction processing replaces flash-affected lines with lines from past screens, then image quality is improved, but processing delay increases

Engineering Contradiction:
Improveimage qualityVSAvoidprocessing delay
Core Design Contradiction:
Manufacturing precisionVSLoss of time

Solution Approach 1:

The patent performs correction processing by replacing flash-affected lines with corresponding lines from previously captured screens. This preliminary action approach maintains image quality by using already-captured valid line data, while minimizing processing delay since the replacement data is readily available from the buffer without requiring additional capture or complex reconstruction.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP2629508B1Image processing device, image processing method, image processing program for flash detection
Publication Date: 2019.03.27 JVC KENWOOD CORP
  • EP2629508B1 patent drawingFigure 1
  • EP2629508B1 patent drawingFigure 2(a)~2(d)
  • EP2629508B1 patent drawingFigure 3

AI summary

A flash detection unit calculates a line average luminance of each line of the current screen of image data and a screen average luminance of a past screen at least one screen before the current screen and compares the calculated line average luminance with the calculated screen average luminance to detect whether the current screen includes a line of high luminance due to a flash. A holding unit holds the past screen of the image data. A flash correction unit, if it is detected that some lines of the current screen have high luminance, replaces the lines having high luminance in the current screen with corresponding lines of the past screen held in the holding unit to correct the image data.