Image Sensor Fixed Pattern Noise Compensation via Temperature-Adjusted Dark Current

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for compensating fixed pattern noise (FPN) in image sensors, such as CMOS sensors, face challenges like temperature dependence, increased transistor requirements, and inefficiencies in measurement and correction processes, which affect the accuracy and efficiency of noise reduction.

Innovation Solution

A method that measures reference dark currents at a reference temperature and applies a temperature-dependent adjustment factor to correct for dark current variations, allowing for FPN compensation without additional time or chip area, using a darkened pixel as a temperature sensor and varying integration and storage times to determine reference values through regression analysis.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a dark frame is subtracted from the actual image for FPN compensation, then the fixed pattern noise is reduced, but the compensation is only valid for the operating point at which the dark frame was measured and becomes inaccurate at different operating points due to temperature dependence

Engineering Contradiction:
ImproveFPN compensation accuracyVSAvoidtemperature adaptability
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent applies parameter changes by measuring reference dark currents at multiple different temperatures and using temperature as a variable parameter to select or interpolate the appropriate compensation values. The system stores reference dark currents measured at various temperatures and uses the current temperature to determine which reference values to apply, thereby adapting the FPN compensation to different operating temperatures and maintaining accuracy across varying conditions.

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If Correlated Double Sampling is used for FPN compensation, then the fixed pattern noise is reduced, but more transistors are required per pixel which reduces the usable chip area for photosensitive areas

Engineering Contradiction:
ImproveFPN compensation accuracyVSAvoidusable chip area
Core Design Contradiction:
Measurement precisionVSArea of stationary object

Solution Approach 1:

The patent extracts the FPN compensation function from the pixel-level circuit operations (like CDS) and moves it to a post-processing stage. Instead of requiring additional transistors within each pixel for sampling operations, the system captures a single dark frame, measures its statistics externally, and uses these measurements to compensate FPN in subsequent images through software processing. This extracts the compensation functionality from the physical pixel structure, preserving chip area for photosensitive elements.

Inventive Principle:
Principle #2Taking out (Extraction)

3Measurement precision

If a dark frame is updated periodically or in response to temperature change for FPN compensation, then the compensation accuracy is maintained, but operating time is lost while the camera generates the current dark frame

Engineering Contradiction:
ImproveFPN compensation accuracyVSAvoidoperating time
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent applies preliminary action by performing comprehensive FPN compensation measurements at multiple temperatures in advance, before actual operation. The system pre-measures reference dark currents at various temperatures and stores these results for later use. During operation, the system simply retrieves and applies the appropriate pre-measured compensation values based on the current temperature, eliminating the need for time-consuming dark frame updates during productive operation while maintaining high compensation accuracy.

Inventive Principle:
Principle #10Preliminary action

4Measurement precision

If FPN compensation is performed for each pixel individually, then the compensation accuracy is improved, but the measurement and correction effort increases significantly

Engineering Contradiction:
Improvepixel-level compensation accuracyVSAvoidmeasurement and correction effort
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent merges the measurement process by capturing a single dark frame that contains information from all pixels simultaneously, rather than measuring each pixel separately. The system then processes this unified dark frame to extract FPN characteristics for all pixels in one operation. This combining approach maintains pixel-level compensation accuracy while dramatically reducing measurement and correction effort compared to individual pixel measurements.

Inventive Principle:
Principle #5Merging (Combining)

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

This approach enables effective FPN compensation, particularly for Dark Signal Non-Uniformity (DSNU), across various operating states without increasing measurement effort or chip area, reducing the need for frequent temperature-dependent measurements and improving noise reduction efficiency.

Implementation Method 1

Each pixel therefore has a specific error. If this error is known, it can be corrected.

Methodology Applied
Scientific EffectPhotoelectric effect: Photoelectric Effect

Implementation Method 2

a temperature sensor (18) is provided for measuring a current temperature T of the image sensor (12)

Methodology Applied
Scientific EffectTemperature sensing: Thermocouple

Data Source

PatentEP3591962B1Compensation of fixed pattern noise of a image sensor
Publication Date: 2020.09.09 SICK AG
  • EP3591962B1 patent drawingFigure 1~2
  • EP3591962B1 patent drawingFigure 3~4
  • EP3591962B1 patent drawing

AI summary

A method for compensating fixed-pattern noise in an image sensor (12) with a plurality of pixels (20) is described, in which the signal of each pixel (20) is corrected based on a dark current determined for that pixel (20). The temperature T of the image sensor (12) is measured, and the dark current of each pixel (20) is determined by adjusting a reference dark current (DCRef,PD, DCRef,Sp), previously determined for that pixel (20) at a reference temperature TRef, to the temperature T using a temperature-dependent adjustment factor.