Stereo Camera Sensor Verification via Image Comparison

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for checking the functionality of stereo camera image sensors are inadequate, as they fail to reliably detect errors such as stuck-at pixels, address decoder errors, and contamination, especially in static scenarios and single-channel architectures, which can lead to safety failures in security applications.

Innovation Solution

A method that compares the recorded images from two image sensors with a perspective offset using a stereo algorithm to estimate disparities, allowing for the identification of matching image sections and detecting errors by analyzing differences in pixel values and image features, which is integrated into the depth map calculation for real-time functionality checking.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If current noise level comparison method is used to check image sensor functionality, then the test can be performed, but the detection reliability is insufficient because defective pixels can show noise corresponding to expectations

Engineering Contradiction:
Improvedetection reliabilityVSAvoidtest method complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent uses the stereo camera's dual-camera architecture to create a copy of the scene from a slightly different perspective. By comparing corresponding pixels between the two images, the system can detect defective pixels that would otherwise be indistinguishable from normal noise in a single-image test.

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The system implements continuous monitoring of pixel correspondence between stereo images, providing real-time feedback on image sensor health. When discrepancies exceed thresholds, the system can trigger alerts or corrective actions, enabling proactive maintenance.

Inventive Principle:
Principle #23Feedback

2Reliability

If FPN mixed-up detection method is used to detect address decoder errors, then address errors can be detected, but the FPN cannot be evaluated with sufficient reliability because it is overlaid by the actual image

Engineering Contradiction:
Improveerror detection reliabilityVSAvoidFPN evaluation difficulty
Core Design Contradiction:
ReliabilityVSDifficulty of detecting and measuring

Solution Approach 1:

The stereo camera captures a duplicate view of the scene, allowing the system to separate fixed pattern noise from actual scene content by comparing the two images. Address decoder errors manifest as consistent mismatches in corresponding pixel locations across the stereo pair.

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The system performs preliminary comparison of stereo image pairs to establish baseline FPN characteristics before overlaying with actual image content, enabling reliable separation and evaluation of FPN components.

Inventive Principle:
Principle #10Preliminary action

3Reliability

If verification circuits are integrated in the image sensor to verify functionality, then functional verification can be performed, but the image sensor becomes more complex to produce and requires additional space

Engineering Contradiction:
Improvefunctional verification capabilityVSAvoidimage sensor complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The stereo camera's dual-camera system serves multiple functions: it captures depth information for 3D imaging while simultaneously providing redundant data for image sensor verification. This eliminates the need for dedicated verification circuits, as the second camera acts as both a functional component and a verification tool.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The stereo camera system performs self-verification by using its own dual-camera architecture to monitor and detect image sensor defects. The system verifies its own functionality without requiring external verification equipment or additional on-sensor circuitry.

Inventive Principle:
Principle #25Self-service

4Reliability

If protective fields are configured to prevent operator intrusion, then safety protection is provided, but the system may trigger false safety stops due to undetected image sensor errors

Engineering Contradiction:
Improvesafety protection reliabilityVSAvoidsafety system complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The continuous comparison of stereo image pairs provides real-time feedback on sensor health, enabling the safety system to distinguish between genuine safety threats and artifacts caused by defective pixels or sensor errors, thereby reducing false positives.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP3496398B1Secure stereo camera and method for testing the functionality of the image sensors
Publication Date: 2020.01.29 SICK AG
  • EP3496398B1 patent drawingFigure 1
  • EP3496398B1 patent drawingFigure 2
  • EP3496398B1 patent drawing

AI summary

A safe stereo camera (10) is described, comprising at least one first image sensor (16a) for capturing a first raw image (36a) and a second image sensor (16b) for capturing a second raw image (36b) of a monitored area (12) from a displaced perspective, a stereoscopic unit (26) for generating a depth map from the two raw images (36a-b), and a test unit (28) for verifying the functionality of the image sensors (16a-b). The test unit (28) is configured to detect errors in the image sensors (16a-b) by comparing first image information from a first image window (40a) of the first raw image (36a) with second image information from a second image window (40b) of the second raw image (36b) and by selecting the image windows (40a-b) based on information from the stereoscopic unit (26) when generating the depth map.