Datasets¶

Typearray

A dataset is a container for data of a homogeneous nature, characterized by standardized data classes, with its volume necessitating storage within the HDF5 structure. Through the process it originates, the dataset establishes a connection between the data and physical reality (time/space).

The datasets array describes datasets properties and how they relate to the acquisition and post-processing processes.

Property	Type	Technology	Description
id	integer	All	The unique dataset id within the group
name	string	All	The name of the dataset
storageMode	string	UT	One of `Independent` or `Paintbrush`. See the storageMode conventions.
dataTransformations	array	All	A dataTransformations array describing the various operations to perform on the dataset
path	string	All	The path to the dataset in the HDF5 structure
dataClass	string	All	UT One of: `AScanAmplitude`, `AScanStatus`, `TfmValue`, `TfmStatus`, `FiringSource`, `CScanPeak`, `CScanStatus`, `CScanTime` ET One of: `Impedance`, `ImpedanceStatus`
dataValue `required`	object	All	A dataValue object
dimensions `required`	array	All	A dimensions array. See the supported datasets section for the possible configurations.

dataTransformations¶

Typearray

The dataTransformations array references the last process of the processes chain to consider for all transformations that should be applied to the dataset. See the typical groups structure examples.

Property	Type	Description
processId `required`	integer	The processId of the last process in the chain
groupId	integer	The groupId of the last process in the chain (not required if in the same group)

Example

"dataTransformations": [
  {
    "processId": 1
  }
]

dataValue¶

Typeobject

The dataValue object describes the range or possible values a dataset can contain.

UT For AScanAmplitude, TfmValue, CScanPeak, and CScanTime data classes

ET For Impedance data class

Property	Type	Description
min `required`	number	Minimum value possible of the dataset
max `required`	number	Maximum value possible of the dataset
unitMin `required`	integer	Minimum value possible of the dataset in the corresponding unit
unitMax `required`	integer	Maximum value possible of the dataset in the corresponding unit
unit `required`	string	UT One of the following: `Percent`, `Coherence`, `Seconds` ET One of the following: `Percent`

To map a value from the range [min, max] to [unitMin, unitMax], use the following formula:

\[ x_{scaled}= \left( \frac{x - \text{min}}{\text{max} - \text{min}} \right) \times (\text{unitMax} - \text{unitMin}) + \text{unitMin} \]

Info

ET Eddy current impedance data is stored as a complex number. The HDF5 dataset uses a compound type with an r key for the real component and an i key for the imaginary component. The min, max, unitMin, and unitMax scaling is applied independently to each component.

Example

"dataValue": {
  "min": 0,
  "max": 32767,
  "unitMin": 0.0,
  "unitMax": 200.0,
  "unit": "Percent"
}

UT For AScanStatus, TfmStatus, and CScanStatus data classes

Property	Type	Description
hasData `required`	integer	Has data status value
noSynchro `required`	integer	No synchronization status value
saturated `required`	integer	Saturated status value
unit `required`	string	`Bitfield`

ET For ImpedanceStatus data class

Property	Type	Description
hasData `required`	integer	Has data status value
saturated `required`	integer	Saturated status value
unit `required`	string	`Bitfield`

Note

A bitfield array is a compact way to store multiple flags using only a few bits per item. Each flag is read or written using bitwise operations on the stored integer value.

UT exampleET example

"dataValue": {
  "hasData": 1, //(1)!
  "saturated": 2, //(2)!
  "noSynchro": 4, //(3)!
  "unit": "Bitfield"
}

hasData is encoded on the first bit (binary 001 → decimal 1)
saturated is encoded on the second bit (binary 010 → decimal 2)
noSynchro is encoded on the third bit (binary 100 → decimal 4)

So hasData + saturated evaluates to:

Flags	Binary	Decimal
hasData + saturated	011	3

"dataValue": {
  "hasData": 1, //(1)!
  "saturated": 2, //(2)!
  "unit": "Bitfield"
}

hasData is encoded on the first bit (binary 01 → decimal 1)
saturated is encoded on the second bit (binary 10 → decimal 2)

dimensions¶

Typearray

The dimensions array describes the different dimensions (axes) of the dataset(s).

Important

Dimensions are always given in the same order as the HDF5 dataset dimensions.

UT For UCoordinate, VCoordinate, WCoordinate, Ultrasound, and StackedAScan axes:

Property	Type	Unit	Description
axis `required`	string	-	One of the following: `UCoordinate`, `VCoordinate`, `WCoordinate`, `Ultrasound`,`StackedAScan`
offset	number	m or s	Offset to position the dataset
quantity `required`	integer	-	Size of the dataset against this axis (dimension)
resolution `required`	number	m or s	Resolution of the dataset against this axis (dimension)

Example

"dimensions": [
  {
    "axis": "UCoordinate",
    "offset": 0.0,
    "quantity": 101,
    "resolution": 0.001
  },
  {
    "axis": "VCoordinate",
    "offset": 0.0,
    "quantity": 57,
    "resolution": 0.001
  },
  {
    "axis": "Ultrasound",
    "offset": -1.01E-06,
    "quantity": 364,
    "resolution": 2E-08
  }
]

UT For the Beam axis:

Property	Type	Description
axis `required`	string	`Beam`
beams `required`	array	A beams array

beams¶

UT Typearray

Property	Type	Unit	Description
id	integer	-	The unique beam id within the acquisition process
velocity `required`	number	m/s	The velocity used to calculate focal laws
skewAngle `required`	number	°	The beam skew angle
refractedAngle `required`	number	°	The beam refracted angle in the specimen
uCoordinateOffset `required`	number	m	The beam exit point offset w.r.t to \(u\) axis
vCoordinateOffset `required`	number	m	The beam exit point offset w.r.t to \(v\) axis
ultrasoundOffset `required`	number	s	The beam offset w.r.t to time (ultrasound) axis

Example

"dimensions": [
  {
    "axis": "UCoordinate",
    "quantity": 301,
    "resolution": 0.001
  },
  {
    "axis": "Beam",
    "beams": [
      {
        "velocity": 3240.0,
        "skewAngle": 90.0,
        "refractedAngle": 40.0,
        "uCoordinateOffset": 0.0,
        "vCoordinateOffset": -0.03868746281893342,
        "ultrasoundOffset": 0.0
      },
      {
        "velocity": 3240.0,
        "skewAngle": 90.0,
        "refractedAngle": 41.0,
        "uCoordinateOffset": 0.0,
        "vCoordinateOffset": -0.038554784991032057,
        "ultrasoundOffset": 0.0
      },
      {...}]},
  {
    "axis": "Ultrasound",
    "quantity": 1700,
    "resolution": 2E-08
  }
]

Note: The uCoordinateOffset and vCoordinateOffset are only applicable when the surface of the wedge matches the specimen surface on which offsets are referenced.

ET For Channel and AcquisitionCycle axes:

Property	Type	Description
axis `required`	string	One of the following: `Channel`, `AcquisitionCycle`
quantity `required`	integer	Size of the dataset against this axis (dimension)

Example

"dimensions": [
  {
    "axis": "Channel",
    "quantity": 16
  },
  {
    "axis": "AcquisitionCycle",
    "quantity": 200
  }
]

These dimensions can vary depending on the group and scan types.

Supported datasets¶

The datasets shape may vary depending on the type of scanning and data mapping strategy. The following types are defined:

One Line Scanning: The probe is mechanically moved against a single axis on the surface of the specimen.
Raster Scanning: The probe is mechanically moved against two axes on the surface of the specimen.
ET All Cycle: Data is stored cycle by cycle using an allCycle data mapping, without prior spatial indexing. A subsequent mapToDescrete process converts it to a spatially-indexed grid.

Conventional ultrasonic testing (UT)¶

For conventional ultrasonic testing (UT), two datasets are supported: one containing the raw A-Scans (dataClass = AScanAmplitude), and another containing the status information (dataClass = AScanStatus, see above) for each A-Scan. The dimensions of each dataset can be organized as illustrated below.

Example files: Manual weld scanning

Phased array ultrasonic testing (PAUT)¶

For phased array ultrasonic testing (PAUT), two datasets are supported: one containing the raw A-Scans (dataClass = AScanAmplitude), and another containing the status information (dataClass = AScanStatus, see above) for each A-Scan. The dimensions of each dataset can be organized as illustrated below. Three cases are presented, depending on the type of scanning and storage mode.

One Line Scanning0° Raster ScanningAngle Beam Raster Scanning

Example files: Manual weld scanning, Semiautomated weld scanning

Note that an additional dataset with dataClass = FiringSource is also used in this configuration but not documented to date.

Example files: Composite wheel probe scanning, Corrosion inspection, Pipe elbow corrosion

Example files: Composite wheel probe scanning

Total focusing method and its derivatives (TFM/PCI/PWI)¶

For total focusing method and its derivatives (TFM/PCI/PWI), two datasets are supported: one containing the reconstructed images (dataClass = TfmValue), and another containing the status information (dataClass = TfmStatus, see above) for each image. The dimensions of each dataset can be organized as illustrated below.

Example files: Manual weld scanning, Girth weld scanning, Composite X-Y scanning

Full Matrix Capture (FMC)¶

For Full Matrix Capture (FMC), one dataset type is supported which contains the elementary A-Scans (dataClass = AScanAmplitude). The dimensions of the dataset can be organized as illustrated below.

The stackedAScan dimension is organized as illustrated below:

Example file: Full matrix capture

Conventional Eddy Current Testing (ECT) and Eddy Current Array (ECA)¶

For Eddy Current Testing (ECT) and Eddy Current Array (ECA), the supported dataset type contains the raw impedance strip charts (dataClass = Impedance). The dimensions of the dataset can be organized as illustrated below.