Adaptive wavelet neural network for terrestrial laser scanner-based crack detection

Türkan, Yelda; Hong, Jonathan; Laflamme, Simon; Puri, Nisha

doi:10.1016/j.autcon.2018.06.017

Cited by 74 publications

(30 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…As a result, numerous studies have evaluated machine learning approaches to detect cracks from planar surfaces. In one recent example, Turkan et al developed an adaptive wavelet neural network-based approach for point cloud data to detect cracks from concrete surfaces [16]. Within this study, the developed network accepts the X and Y components of point clouds and estimates Z coordinates to localize and characterize anomalies.…”

Section: Damage Detection Based On Geometrical Featuresmentioning

confidence: 99%

Non-Temporal Point Cloud Analysis for Surface Damage in Civil Structures

Mohammadi

Wood

Wittich

2019

IJGI

View full text Add to dashboard Cite

Assessment and evaluation of damage in civil infrastructure is most often conducted visually, despite its subjectivity and qualitative nature in locating and verifying damaged areas. This study aims to present a new workflow to analyze non-temporal point clouds to objectively identify surface damage, defects, cracks, and other anomalies based solely on geometric surface descriptors that are irrespective of point clouds’ underlying geometry. Non-temporal, in this case, refers to a single dataset, which is not relying on a change detection approach. The developed method utilizes vertex normal, surface variation, and curvature as three distinct surface descriptors to locate the likely damaged areas. Two synthetic datasets with planar and cylindrical geometries with known ground truth damage were created and used to test the developed workflow. In addition, the developed method was further validated on three real-world point cloud datasets using lidar and structure-from-motion techniques, which represented different underlying geometries and exhibited varying severity and mechanisms of damage. The analysis of the synthetic datasets demonstrated the robustness of the proposed damage detection method to classify vertices as surface damage with high recall and precision rates and a low false-positive rate. The real-world datasets illustrated the scalability of the damage detection method and its ability to classify areas as damaged and undamaged at the centimeter level. Moreover, the output classification of the damage detection method automatically bins the damaged vertices into different confidence intervals for further classification of detected likely damaged areas. Moving forward, the presented workflow can be used to bolster structural inspections by reducing subjectivity, enhancing reliability, and improving quantification in surface-evident damage.

show abstract

Section: Damage Detection Based On Geometrical Featuresmentioning

confidence: 99%

Non-Temporal Point Cloud Analysis for Surface Damage in Civil Structures

Mohammadi

Wood

Wittich

2019

IJGI

View full text Add to dashboard Cite

show abstract

“…Single-frame fringe analysis techniques only need one frame of fringe pattern to extract phase information so it is more suitable to high speed measurement. It mainly contains Fourier Transform profilometry (FTP) [9][10][11], Windowed Fourier Transform profilometry(WFTP) [12][13][14], Wavelet Transform profilometry (WTP) [15][16][17][18] and S-Transform profilometry (STP) [19][20][21][22]. However, there are still shortcomings for these methods.…”

Section: Introductionmentioning

confidence: 99%

“…WFTP [12]makes an analysis of a fringe pattern by a size-fixed window so it can not get accurate results of the tested surface with sharp change. The WTP [16] calculates the correlation between the deformed fringe pattern and the wavelets with the different dilation values to extract the phase. In the wavelet transform, the scale factor instead of the frequency concept is employed to analyze the fringe pattern.…”

Section: Introductionmentioning

confidence: 99%

Fast 3D Shape Measurement of Complex Objects Combining S-transform Profilometry and Sub-Pixel Refinement

Yang

2020

Preprint

View full text Add to dashboard Cite

Abstract In the fast 3D shape measurement of complex objects, the process of phase unwrapping affects the measurement accuracy and measurement speed. To omit phase unwrapping, a fast 3D shape measurement method is proposed based on S-transform profilometry and sub-pixel refinement. In this work, Deformed fringe pattern is analyzed based on S-transform to get wrapped phase. To reduce the effect of phase unwrapping, a stereo vision system is employed. Firstly, a deformed fringe pattern is directly used and a hybrid function is proposed to get an initial parallax. Secondly, a phase slope algorithm is proposed to confirm edge point. At last, sub-pixel parallax is refined fromIcoordinate. Experiments are carried out to verify the feasibility and advantage of the proposed method. Results verify that the proposed method can provide fast and accurate 3D shape measurement of complex objects, especially for discontinuous objects and shiny objects.

show abstract

“…For example, in the case of wind turbine generators, about 25 % to 30 % of the overall wind power generation costs account for operation and maintenance, while wind turbine blade failure results in more than 7 days of downtime [1]. The approach of SHM and CM exploits collection and analysis of signals continuously measured with mounted sensors or visual images captured with aerial vehicles [1] or laser scanners [2].…”

Section: Introductionmentioning

confidence: 99%

“…The most popular technique for timefrequency analysis is wavelet transform, which can be used for both stationary and non-stationary signals [3]. This method can also be effectively applied to structural damage interrogation composite structures to track delamination damage [5;6] and concrete structures to monitor crack propagation [2] among other cases.…”

Section: Introductionmentioning

confidence: 99%

Review on time-frequency-based machine learning for structural damage assessment and condition monitoring

Janeliukštis

2019

Engineering for Rural Development

View full text Add to dashboard Cite

Safe operation of complex structures and machines cannot be achieved by visual inspections alone. Thus, simple, efficient and cost-effective inspection methods are of paramount importance. This review article is aimed at providing an insight into approaches of health monitoring and condition monitoring of structures and machines based on machine learning. Machine learning is an approach of using data to construct predictive models giving predictions of existing and future trends. In the case of structural health monitoring and condition monitoring of machinery, it is able to detect damage and distinguish between different damage types and severities, thus being able to provide early warnings of failure of structural and machinery components. In this review paper, the focus is on information obtained from time-frequency features, which provides an advantage with respect to features in time domain and frequency domain of tracking changes of structural integrity in time, which is especially useful in dealing with non-stationary signals. Details of damage-sensitive feature extraction, data analysis and decision making are provided.

show abstract

Adaptive wavelet neural network for terrestrial laser scanner-based crack detection

Cited by 74 publications

References 47 publications

Non-Temporal Point Cloud Analysis for Surface Damage in Civil Structures

Non-Temporal Point Cloud Analysis for Surface Damage in Civil Structures

Fast 3D Shape Measurement of Complex Objects Combining S-transform Profilometry and Sub-Pixel Refinement

Review on time-frequency-based machine learning for structural damage assessment and condition monitoring

Contact Info

Product

Resources

About