Instrumentation & Measurement Magazine 25-2 - 12

Sparse tensor model is more suitable for video stream data
processing than sparse matrix model, so it has been widely
used in ECPT sequence post-processing recently. In addition to
TRPCA (decompose a tensor with the form of sum), there are
CP decomposition and Tucker decomposition, which decompose
a tensor with the product form of tensors and matrices.
These different decomposition methods enrich the ECPT sequences
post-processing algorithms.
At present, effect and efficiency are the two main research
focuses of sparse decomposition algorithms in ECPT. Effect is
the devotion to decompose the defect information in more detail.
The main research ideas on this area can be summarized
as:
◗ Modify the objective function of optimization to describe
the components of the raw data more accurately. For
example, [21] added a noise component N into the standard
sparse decomposition model and measured it by F
norm to reduce the noise in sparse component.
◗ Extract as many sparse components as possible through
multi-layer structure decomposition. For example, [22]
and [23] designed the structure of multi-layer sparse
decomposition to perform sparse decomposition again
for the non-sparse components in previous decomposition
results.
Efficiency, as the name suggests, is the devotion to reduce
the run-time of the sparse decomposition algorithms. Because
the sparse decomposition contains a large number of matrix
or tensor product operations, the time complexity is higher
than aforementioned transform domain algorithms and statistic
algorithms. In computer vision (CV) and optimization,
many scholars reduce the complexity of the operation of the
objective function through mathematical techniques. For example,
in the iteration of aforementioned TNN-TRPCA, Fast
Fourier Transform is used many times to accelerate the calculation
of optimization. This research is more oriented towards
general scenarios rather than ECPT sequences specifically.
Recently, some scholars have tried to improve the efficiency
by combining the unique characteristics of ECPT sequences.
For example, based on the spatial cohesion feature of defects
in ECPT sequences, [1] introduced Moran's index as a criterion
for early stop iteration. Compared with the standard TRPCA
method, the number of iterations can be reduced by 82% to
97%, and the overfitting can also be improved. This kind of
work promotes the practical application of sparse decomposition
in ECPT, which is a worthy research direction that should
be given attention in the future.
Summary
With development of a few decades, the variety of ECPT
sequence post-processing algorithms has been gradually enriched.
From the earliest FFT, TSR, ICA, to the recent sparse
model decomposition, rich post-processing algorithms
promote the further application of ECPT. Table 2 summarizes
the algorithms mentioned above. In fact, each type of
post-processing algorithm has its own advantages and disadvantages.
In some cases, some traditional methods are
even more proper than newly developed algorithms. Therefore,
in the industrial measurement, we should choose the
post-processing algorithm according to the needs. It is worth
mentioning that, although FFT and ICA still have a place in
engineering, with the advent of artificial intelligence, laboratory
studies have tended toward sparse decomposition
algorithms, neural networks and the more advanced data
analysis methods, even if there is a certain distance from lab
to engineering.
Recommendations for Further Research
ECPT has been developed rapidly in theoretical models, hardware
equipment and data processing algorithms recently.
However, this technology still faces some challenges to be improved.
To sum up, the following three aspects on ECPT are
worth further exploration.
Type
Transform domain
FFT
TSR
Table 2 - Summary of the aforementioned algorithms
Advantages
Fastest processing speed
Disadvantages
Poor Robustness on complex specimens
Binary segmentation
Statistical
ICA
Sparse
decomposition
12
Matrix decomposition
Tensor decomposition
Identify multi different
components, better
highlight defect
information
Highlight defect information
best, strong robustness
Fast processing speed
Temperature sensitive only, single frame
processing only, false alarm on nondefective
areas with high temperature
(coil, specimen edge), unable to identify
illusory temperature caused by uneven
reflectivity
The number of independent components
needs to be specified in advance, the
component corresponding to defect
information needs secondary screening
Highest time cost, high requirement on
computer configuration, sometimes
overfitting in industrial application
IEEE Instrumentation & Measurement Magazine
April 2022

Instrumentation & Measurement Magazine 25-2

Table of Contents for the Digital Edition of Instrumentation & Measurement Magazine 25-2

Instrumentation & Measurement Magazine 25-2 - Cover1
Instrumentation & Measurement Magazine 25-2 - Cover2
Instrumentation & Measurement Magazine 25-2 - 1
Instrumentation & Measurement Magazine 25-2 - 2
Instrumentation & Measurement Magazine 25-2 - 3
Instrumentation & Measurement Magazine 25-2 - 4
Instrumentation & Measurement Magazine 25-2 - 5
Instrumentation & Measurement Magazine 25-2 - 6
Instrumentation & Measurement Magazine 25-2 - 7
Instrumentation & Measurement Magazine 25-2 - 8
Instrumentation & Measurement Magazine 25-2 - 9
Instrumentation & Measurement Magazine 25-2 - 10
Instrumentation & Measurement Magazine 25-2 - 11
Instrumentation & Measurement Magazine 25-2 - 12
Instrumentation & Measurement Magazine 25-2 - 13
Instrumentation & Measurement Magazine 25-2 - 14
Instrumentation & Measurement Magazine 25-2 - 15
Instrumentation & Measurement Magazine 25-2 - 16
Instrumentation & Measurement Magazine 25-2 - 17
Instrumentation & Measurement Magazine 25-2 - 18
Instrumentation & Measurement Magazine 25-2 - 19
Instrumentation & Measurement Magazine 25-2 - 20
Instrumentation & Measurement Magazine 25-2 - 21
Instrumentation & Measurement Magazine 25-2 - 22
Instrumentation & Measurement Magazine 25-2 - 23
Instrumentation & Measurement Magazine 25-2 - 24
Instrumentation & Measurement Magazine 25-2 - 25
Instrumentation & Measurement Magazine 25-2 - 26
Instrumentation & Measurement Magazine 25-2 - 27
Instrumentation & Measurement Magazine 25-2 - 28
Instrumentation & Measurement Magazine 25-2 - 29
Instrumentation & Measurement Magazine 25-2 - 30
Instrumentation & Measurement Magazine 25-2 - 31
Instrumentation & Measurement Magazine 25-2 - 32
Instrumentation & Measurement Magazine 25-2 - 33
Instrumentation & Measurement Magazine 25-2 - 34
Instrumentation & Measurement Magazine 25-2 - 35
Instrumentation & Measurement Magazine 25-2 - 36
Instrumentation & Measurement Magazine 25-2 - 37
Instrumentation & Measurement Magazine 25-2 - 38
Instrumentation & Measurement Magazine 25-2 - 39
Instrumentation & Measurement Magazine 25-2 - 40
Instrumentation & Measurement Magazine 25-2 - 41
Instrumentation & Measurement Magazine 25-2 - 42
Instrumentation & Measurement Magazine 25-2 - 43
Instrumentation & Measurement Magazine 25-2 - 44
Instrumentation & Measurement Magazine 25-2 - 45
Instrumentation & Measurement Magazine 25-2 - 46
Instrumentation & Measurement Magazine 25-2 - 47
Instrumentation & Measurement Magazine 25-2 - 48
Instrumentation & Measurement Magazine 25-2 - 49
Instrumentation & Measurement Magazine 25-2 - 50
Instrumentation & Measurement Magazine 25-2 - 51
Instrumentation & Measurement Magazine 25-2 - 52
Instrumentation & Measurement Magazine 25-2 - 53
Instrumentation & Measurement Magazine 25-2 - 54
Instrumentation & Measurement Magazine 25-2 - 55
Instrumentation & Measurement Magazine 25-2 - 56
Instrumentation & Measurement Magazine 25-2 - 57
Instrumentation & Measurement Magazine 25-2 - 58
Instrumentation & Measurement Magazine 25-2 - 59
Instrumentation & Measurement Magazine 25-2 - 60
Instrumentation & Measurement Magazine 25-2 - 61
Instrumentation & Measurement Magazine 25-2 - 62
Instrumentation & Measurement Magazine 25-2 - 63
Instrumentation & Measurement Magazine 25-2 - 64
Instrumentation & Measurement Magazine 25-2 - 65
Instrumentation & Measurement Magazine 25-2 - 66
Instrumentation & Measurement Magazine 25-2 - 67
Instrumentation & Measurement Magazine 25-2 - 68
Instrumentation & Measurement Magazine 25-2 - 69
Instrumentation & Measurement Magazine 25-2 - 70
Instrumentation & Measurement Magazine 25-2 - 71
Instrumentation & Measurement Magazine 25-2 - 72
Instrumentation & Measurement Magazine 25-2 - 73
Instrumentation & Measurement Magazine 25-2 - 74
Instrumentation & Measurement Magazine 25-2 - 75
Instrumentation & Measurement Magazine 25-2 - 76
Instrumentation & Measurement Magazine 25-2 - 77
Instrumentation & Measurement Magazine 25-2 - 78
Instrumentation & Measurement Magazine 25-2 - 79
Instrumentation & Measurement Magazine 25-2 - 80
Instrumentation & Measurement Magazine 25-2 - 81
Instrumentation & Measurement Magazine 25-2 - 82
Instrumentation & Measurement Magazine 25-2 - 83
Instrumentation & Measurement Magazine 25-2 - 84
Instrumentation & Measurement Magazine 25-2 - Cover3
Instrumentation & Measurement Magazine 25-2 - Cover4
https://www.nxtbook.com/allen/iamm/26-6
https://www.nxtbook.com/allen/iamm/26-5
https://www.nxtbook.com/allen/iamm/26-4
https://www.nxtbook.com/allen/iamm/26-3
https://www.nxtbook.com/allen/iamm/26-2
https://www.nxtbook.com/allen/iamm/26-1
https://www.nxtbook.com/allen/iamm/25-9
https://www.nxtbook.com/allen/iamm/25-8
https://www.nxtbook.com/allen/iamm/25-7
https://www.nxtbook.com/allen/iamm/25-6
https://www.nxtbook.com/allen/iamm/25-5
https://www.nxtbook.com/allen/iamm/25-4
https://www.nxtbook.com/allen/iamm/25-3
https://www.nxtbook.com/allen/iamm/instrumentation-measurement-magazine-25-2
https://www.nxtbook.com/allen/iamm/25-1
https://www.nxtbook.com/allen/iamm/24-9
https://www.nxtbook.com/allen/iamm/24-7
https://www.nxtbook.com/allen/iamm/24-8
https://www.nxtbook.com/allen/iamm/24-6
https://www.nxtbook.com/allen/iamm/24-5
https://www.nxtbook.com/allen/iamm/24-4
https://www.nxtbook.com/allen/iamm/24-3
https://www.nxtbook.com/allen/iamm/24-2
https://www.nxtbook.com/allen/iamm/24-1
https://www.nxtbook.com/allen/iamm/23-9
https://www.nxtbook.com/allen/iamm/23-8
https://www.nxtbook.com/allen/iamm/23-6
https://www.nxtbook.com/allen/iamm/23-5
https://www.nxtbook.com/allen/iamm/23-2
https://www.nxtbook.com/allen/iamm/23-3
https://www.nxtbook.com/allen/iamm/23-4
https://www.nxtbookmedia.com