Contactless Surface Measurement With Motion-Blur Deconvolution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing optical measurement methods using measuring cameras for large workpieces are time-consuming due to the need for multiple image captures and suffer from measurement inaccuracies caused by image artifacts resulting from relative motion between the camera and the object.
Innovation Solution
A method involving continuous relative motion between the measuring camera and the object's surface, capturing multiple images with varying acquisition parameters to correct blurring using deconvolution algorithms with differently structured convolution kernels, and a device comprising a traversing mechanism and evaluation unit for image correction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the measuring camera is moved rapidly to reduce measurement time, then productivity is improved, but measurement precision deteriorates due to motion blur and image artifacts
Solution Approach 1:
The patent converts the harmful motion blur caused by rapid camera movement into a beneficial effect by using deconvolution algorithms. The motion blur contains information about the camera's movement, which is used to construct convolution kernels that, when reversed through deconvolution, sharpen the images and remove artifacts while maintaining the benefits of fast scanning
Solution Approach 2:
The patent replaces the mechanical requirement of complete camera stops with a computational approach. Instead of mechanically stopping the camera to capture sharp images (which reduces productivity), the system uses optical-mechanical continuous movement combined with digital image processing to achieve both speed and precision
2Measurement precision
If multiple images are captured at different positions to cover large surfaces, then measurement precision is improved, but loss of time increases due to repeated stopping and starting
Solution Approach 1:
The patent implements continuous useful action by eliminating the stopping phase between image captures. The measuring camera maintains continuous motion while capturing images at multiple positions, and the deconvolution process handles the motion blur, thereby maintaining surface coverage accuracy without the time loss associated with repeated stops and starts
3Measurement precision
If the camera stops completely to capture sharp images, then measurement precision is improved, but productivity deteriorates due to decay time required for vibrations to subside
Solution Approach 1:
The patent converts the harmful vibration-induced blur into useful information by incorporating it into the convolution kernel model. The vibrations cause predictable motion blur that can be characterized and reversed through deconvolution, allowing sharp images to be obtained without complete stops and eliminating the productivity-reducing decay time
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
Significantly reduces measurement time while maintaining high accuracy by minimizing image artifacts through optimized image acquisition and correction.
Implementation Method 1
optical sensors capture this information using light
Implementation Method 2
blurring of the images caused by the relative motion is corrected using a deconvolution algorithm
Data Source
Figure 1a~3c
Figure 4~7
Figure 5
AI summary
In a method for the non-contact measurement of an object (26) using a measuring camera (24), a continuous relative motion is generated between the measuring camera (24) and a surface (36) of the object (26). During the relative motion, several images of the surface (36) of the object (26) are recorded, each image showing a different section (34) of the surface (36). Any blurring of the images caused by the relative motion is corrected using a deconvolution algorithm, employing different convolution kernels that differ from each other at least at one zero. For example, exactly one image can be recorded of each section of the surface (36), whereby at least one recording parameter is changed during the recording of this single image such that the convolution kernel changes at least at one zero during the recording.