Vertical Color Image Sensor Defect Classification and Correction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for detecting and classifying defects in CMOS color image sensors, such as pixel clusters, chip-surface defects, and window defects, lack effective classification capabilities, leading to high manufacturing costs and yield loss due to the inability to distinguish between different types of defects and correct them efficiently.

Innovation Solution

A method that utilizes all color channel information to classify and correct defects in vertical color pixel image sensors by characterizing each defective pixel, distinguishing between pixel clusters, chip-surface defects, and window defects through multiple characteristics, and implementing a two-pass methodology for defect detection and correction.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional defect testing methods are used that only measure size and quantity of defects, then the testing process is simple and fast, but the classification capability is minimal and yield enhancement is limited

Engineering Contradiction:
Improvedefect classification capabilityVSAvoidtesting process complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the defect analysis process into two distinct passes: a first pass that performs comprehensive multi-characteristic analysis for classification, and a second pass that performs faster verification. This segmentation allows the system to achieve high classification precision while managing complexity by separating detailed analysis from routine verification.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary defect classification in the first pass using multiple characteristics (color channel information, response levels, image type performance) before final verification. This preliminary action enables the system to identify and categorize defects early, allowing for targeted correction strategies and reducing the need for complex repeated analysis.

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If comprehensive multi-characteristic analysis is performed to accurately classify defects, then classification accuracy improves, but calibration time increases

Engineering Contradiction:
Improvedefect classification accuracyVSAvoidcalibration time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent implements periodic action through its two-pass methodology: the first pass performs comprehensive multi-characteristic analysis at specific intervals (during initial calibration) to establish accurate defect classification, while the second pass performs faster verification for routine checks. This periodic comprehensive analysis combined with frequent rapid verification achieves high accuracy while managing calibration time efficiently.

Inventive Principle:
Principle #19Periodic action

Solution Approach 2:

The patent changes analysis parameters between passes: the first pass uses full multi-characteristic analysis (color channels, response levels, image type performance) for comprehensive classification, while the second pass uses simplified verification parameters. This parameter change allows accurate classification when needed while reducing calibration time for routine operations.

Inventive Principle:
Principle #35Parameter changes

3Measurement precision

If all color channel information is utilized for defect classification in vertical color sensors, then classification accuracy improves, but processing complexity increases

Engineering Contradiction:
Improvedefect classification accuracyVSAvoidprocessing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent extracts and utilizes all color channel information (red, green, blue channels) specifically for defect classification purposes in vertical color sensors. By taking out and analyzing the color channel characteristics separately (including stuck channels, offset defects, gain defects), the system achieves high classification accuracy while managing processing complexity through targeted extraction of relevant information rather than processing all sensor data uniformly.

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentEP3706404B1Method of classifying and correcting image sensor defects utilizing defective-pixel information from color channels
Publication Date: 2022.04.06 FOVEON INC
  • EP3706404B1 patent drawingFigure 1
  • EP3706404B1 patent drawingFigure 2
  • EP3706404B1 patent drawingFigure 3

AI summary

A method of classifying and correcting defects in vertical color pixel sensors in utilizing defective pixel information, the method includes defining ranges of output levels of normal pixel sensors from exposure to dark and bright flat field light sources. Dark and bright images are captured, and pixel outputs are measured for each image. Pixels in the dark and bright images having outputs in at least one color channel that are outside of the ranges of the output levels of normal pixel sensors are entered into defective pixel maps. The defective pixels are classified into categories and defective pixel sensors that can be corrected are identified. Correction values for identified defective pixel sensors that can be corrected are generated and entered into a calibration memory associated with the vertical color image sensor.