Contactless Surface Measurement with Motion Deblurring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing contactless measurement methods using measuring cameras are limited by long measurement times and inaccuracies due to image blurring and artefacts from deconvolution processes.
Innovation Solution
A method involving uninterrupted relative motion between the measuring camera and the object, recording multiple images with varying convolution kernels that differ in at least one zero, and applying a deconvolution algorithm to computationally extract blur, using different convolution kernels to prevent artefacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the measuring camera comes completely to rest between successive recordings to avoid image blur, then measurement accuracy is improved, but measurement time increases significantly due to displacement time and decay time
Solution Approach 1:
The patent converts the harmful effect of motion-induced image blur into a beneficial feature by deliberately recording images during uninterrupted relative motion and then using computational deconvolution algorithms to remove the blur. This eliminates the need to stop the measuring camera between recordings, thereby resolving the contradiction between maintaining image sharpness and reducing measurement time
Solution Approach 2:
The patent replaces the mechanical approach of stopping the camera to ensure sharp images with a computational approach. Instead of using mechanical displacement control to prevent blur, the invention uses digital image processing (deconvolution algorithms) to remove blur after capture, substituting mechanical constraints with computational solutions
2Measurement precision
If deconvolution algorithms are applied to deblur motion-blurred images, then measurement accuracy is improved, but artefacts are introduced that lead to measurement inaccuracies
Solution Approach 1:
The patent applies parameter changes by using multiple convolution kernels with different parameters (particularly different zeros in the frequency domain) to perform deconvolution. By varying the kernel parameters and combining results from multiple deconvolution operations, the method reduces artefacts while maintaining image sharpness, thus resolving the contradiction between deblurring and artefact reduction
Solution Approach 2:
The patent uses a composite approach by combining multiple deconvolution results obtained with different convolution kernels. This composite processing method integrates information from multiple computational passes, reducing the impact of artefacts from any single kernel while preserving the deblurred image quality
3Productivity
If faster displacement movements are used between recordings, then measurement time is reduced, but decay time increases due to higher accelerations
Solution Approach 1:
The patent implements continuous useful action by maintaining uninterrupted relative motion between the measuring camera and the object throughout the entire measurement process. The camera never stops moving, and images are captured continuously during motion, eliminating both displacement time and decay time delays, thus resolving the contradiction between measuring speed and decay time
Data Source
AI summary
A method and apparatus are provided to perform contactless measurement of an object with a measuring camera. The method includes producing an uninterrupted relative motion between the measuring camera and a surface of the object. A plurality of images of the surface of the object are recorded during the relative motion, each image showing a different segment of the surface. A blur of the images that is produced by the relative motion is computationally extracted by applying a deconvolution algorithm, use being made of different convolution kernels that differ from one another in at least one zero. In this case, e.g., exactly one image of each segment of the surface can be recorded. During the recording of the exactly one image at least one recording parameter is changed such that the convolution kernel during the recording changes in at least one zero.