| Optical measurement is an advanced technology that combines optoelectronics and mechanical measurement.It involves the non-contact measurement and data processing of objects using optical instruments and corresponding electronic devices.Optical measurement technology offers advantages such as non-destructiveness,high precision,high speed,and non-contact measurement,and finds wide applications in fields such as mechanical engineering,aerospace,optical engineering,manufacturing,and biomedicine.With the development of modern technology,there is a continuous demand for advancements in optical measurement technology.In practical applications,optical measurement techniques require characteristics such as high real-time performance,accuracy,efficiency,and stability.Therefore,researchers need to continuously innovate and improve their technical capabilities to meet the increasing demands for optical measurement technology.However,optical measurement techniques can be subject to various interferences,especially noise interference,which can result in inaccurate measurements and low signal-to-noise ratios,thereby reducing the accuracy and efficiency of optical measurement technology.Consequently,reducing noise in optical measurement objects and improving the signal-to-noise ratio become crucial in addressing this issue.To tackle this problem,researchers have employed digital signal processing techniques,where noise reduction techniques play a vital role.Digital signal processing is a technology that analyzes,processes,and reconstructs signals using computers,with the main objective of improving signal quality and preserving useful information content.In optical measurement,digital signal processing techniques can be used to process raw signals,such as denoising,enhancement,and reconstruction,to improve measurement accuracy and reliability.Among these techniques,noise reduction is an important aspect of digital signal processing.In optical measurement,acquired signals often contain significant amounts of noise due to environmental noise and instrument errors,directly affecting measurement accuracy and reliability.Therefore,noise reduction techniques play a key role in enhancing measurement accuracy and reliability.These techniques can remove noise from corrupted signals,thereby recovering high-quality signals and laying the foundation for subsequent identification,analysis,and decision-making processes.This paper focuses on the application of digital signal processing techniques in the field of optical measurement.It proposes a quaternion wavelet transform(QWT)denoising algorithm applied to different optical measurement scenarios,providing comprehensive theoretical and experimental evidence that combining denoising techniques with existing feature extraction techniques can effectively improve the signal-to-noise ratio and enhance the application effectiveness of optical measurement technology.Additionally,a feedforward neural network algorithm is proposed to extract physical information from optically measured and denoised data.The effectiveness of the proposed approach is verified through comparative experimental studies.The main contributions and innovations of this paper are as follows:1.A denoising algorithm based on the quaternion wavelet transform(QWT)is proposed to process measured images of structured light 3D profiles,enabling accurate measurement in the presence of strong noise.This approach effectively separates useful information and noise in the data while preserving the texture details of the images,achieving precise measurement of structured light 3D profiles under high noise conditions.The proposed approach not only preserves the texture details of the images but also improves their accuracy and timeliness.Compared to traditional algorithms,it offers lower complexity and shorter processing time.Research results demonstrate that in high-noise 3D structured light imaging data,the standard deviation of the original sinusoidal fringe image is reduced from 0.1448 to 0.0192,and the signal-to-noise ratio is improved from 4.6213dB to 13.3463dB.For the measurement of physical masks,the error is less than ±0.02mm,and processing eight frames of 2592×3872 pixel sinusoidal fringe images takes less than 19 seconds.2.A denoising algorithm based on QWT is proposed for processing strong-noise Brillouin gain spectra(BGS)in Brillouin Optical Time Domain Analysis(BOTDA),enabling accurate extraction of Brillouin frequency shift.This approach effectively separates useful information and noise while preserving the detailed information of the signals,improving the accuracy of spatial resolution.Through data processing and comparative experimental studies,the results demonstrate that the proposed approach can process BGS signal images of 200×100,000 points within 32 seconds and achieve a spatial resolution of 3 meters.Furthermore,this approach can be applied to long-distance temperature and stress measurements,offering promising prospects.3.A deep feedforward neural network(FNN)algorithm is proposed to extract environmental temperature data from denoised BGS data processed using the QWT algorithm.Compared to traditional algorithms,this algorithm not only offers fast data extraction and high accuracy but also provides high spatial resolution.It possesses adaptive learning capability,parallel processing capability,fault tolerance,robustness,and efficiency.When the frequency interval is less than 4MHz,the extracted temperature data error remains within±0.11℃,while at 6MHz,it remains within ±0.15℃,with an extraction time of only 17 seconds.This algorithm’s efficiency and accuracy can meet the requirements of health monitoring for many large-scale infrastructure structures.In summary,this paper presents a comprehensive study on optical measurement data analysis and feature extraction methods based on digital signal processing techniques.The proposed approaches,including the QWT denoising algorithm and the FNN algorithm,demonstrate significant advancements in improving the accuracy and effectiveness of optical measurement technology in various scenarios. |
