Monocular Borescope Imaging for Sensor-Free Defect Depth Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Optical instruments lack depth sensors, which are costly and prone to damage in harsh environments, making accurate depth estimation of defects in inaccessible areas challenging, especially in applications like aircraft engines and industrial turbines.
Innovation Solution
A defect depth estimation system using a training system and imaging system that transforms monocular 2D images into a target domain to reduce the domain gap, leveraging machine learning models to estimate defect depth without a depth sensor, utilizing supervised and unsupervised learning techniques to analyze image data from borescopes and other optical instruments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a depth sensor is used to achieve accurate depth estimation, then measurement precision is improved, but device complexity and cost increase
Solution Approach 1:
The patent extracts the depth sensing capability from a dedicated depth sensor and implements it through software-based monocular vision processing. The system removes the need for complex hardware depth sensors by using only standard 2D image sensors combined with machine learning algorithms to estimate depth from single images.
Solution Approach 2:
The patent replaces the mechanical/optical depth sensing system with a computational approach. Instead of using physical depth sensors that require complex hardware, the system substitutes with machine learning models that process 2D images to infer depth information through software-based domain adaptation techniques.
2Measurement precision
If a depth sensor is used to achieve accurate depth estimation, then measurement precision is improved, but reliability decreases due to damage risk in harsh environments
Solution Approach 1:
The patent adopts the philosophy of using robust, simple 2D image sensors that are inherently more durable than complex depth sensors. These standard sensors are less susceptible to damage in harsh environments like aircraft engines, and the system compensates for their limitations through sophisticated software processing rather than relying on fragile hardware.
Solution Approach 2:
The patent replaces the vulnerable mechanical depth sensing system with a computational vision system. By substituting physical depth measurement hardware with software-based depth estimation from standard image sensors, the system achieves both improved reliability in harsh environments and maintained measurement precision through machine learning.
3Measurement precision
If domain adaptation techniques are applied to reduce domain gap, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent applies domain adaptation techniques that transform images from different domains (synthetic training images and real test images) into a common target domain. This involves changing parameters such as lighting conditions, color distributions, and texture characteristics through computational transformations, allowing the machine learning model to generalize accurately across different environments without requiring complex hardware modifications.
Data Source
AI summary
A defect depth estimation system includes a training system and an imaging system that performs defect depth estimation from a monocular 2D image without using a depth sensor. The training system repeatedly receives a first type of image having a defect, and a second type of image that captures the target object having the defect and provides ground truth data indicating an actual depth of the defect. The training system transforms the first domain and the second domain into a target third domain that reduces a domain gap and trains a machine learning model to learn the actual depth of the defect using the target third domain. The imaging system receives a 2D test image in the first forma and uses the trained machine learning model to determine an estimation of the actual depth of the actual defect and to output estimated the estimation of the actual depth.