Deep Style: Using Variational Auto-encoders for Image Generation
Auto Image
-
Upload
nicole-floyd -
Category
Documents
-
view
220 -
download
0
Transcript of Auto Image
-
8/12/2019 Auto Image
1/5
Isothermal Technology Ltd http://www.isotech.co.uk
AUTOMATING TEMPERATURE CALIBRATION BATHS WITH SIMPLE
LOW COST IMAGE ACQUISITION
David J. Southworth
Isothermal Technology Ltd
Abstract
A low cost video camera, Web Cam is used inconjunction with a PC and TemperatureCalibration Bath to automatically calibrate
handheld digital thermometers which have noprovision to be connected to an external computer.
Introduction
Calibration of thermometers including the use of
video cameras is not new. Various NationalMetrology Institutes have used analog and digitalvideo cameras to calibrate Liquid in Glass (LiG)
thermometers with the benefit of recording theinformation and helping to read the liquid columnwithin the thermometers graduations [1]. The
purpose of this paper is to describe a system usinga digital low cost consumer camera to calibratethermometers with a digital display that can only
be recorded manually.
Method
The handheld thermometer is calibrated bycomparing it against a reference probe at a seriesof temperatures. A calibration bath is used to
automatically generate the desired calibrationpoints by computer control. When the system isstable the temperature from the reference or
standard thermometer is recorded to a log file;rather than the operator noting the value from thereadout of the test thermometer a digital camera
captures the display. This allows for unattendedoperation and lower calibration costs.
The software is then able to set the calibrationbath to the next temperature and repeat the
process. At the end of the calibration run a seriesof images are available which show the date andtime the image was created and the images are
captioned with the temperature of the standardprobe.
Equipment
The system consists of
I, A heat source
(Isocal-6 Temperature Calibrator)II, A reference thermometerIII, A PC with appropriate software
IV, A Digital CameraV, To calibrate up to 16 RTDs orthermocouples multiplexers may be used
The heat source is an ISOCAL-6 Calisto model.
The Calisto can be used in different modes. Here itis used a Dry Block Calibrator the referenceprobe and thermometer under test are placed into
suitable pockets of a metal insert which is locatedin the Calistos isothermal calibration volume. Forhigher accuracy requirements it would be possible
to use the Calisto as a Stirred Liquid Bath as theCalisto is a multifunctional heat source withoptions for Dry Block, Liquid Bath, Blackbody and
even ITS-90 Fixed Point Calibration. All theISOCAL-6 models include a serial PC interface formonitoring and control from the PC.
The standard probe is a platinum resistancethermometer which has been calibrated with an
Isotech Model TTI-6 temperature indicator. TheTTI-6 is a high precision portable thermometer it isparticularly suitable as the reference standard for
temperature calibration baths. Based on a highresolution Analogue to Digital converter, allmeasurement computations are performed digitally
without drift. The 5 digit display provides a readoutto 0.01C. The system uncertainty with thereference thermometer over the temperature
-
8/12/2019 Auto Image
2/5
Isothermal Technology Ltd http://www.isotech.co.uk
Set Temperature
Monitor Bath for
Stability
Monitor Standardfor Stability
Acquire ImageLog Reference
Probe + Dry Block
range of the Calisto is 0.025C. The indictor isconnected to the PC via a serial interface.
Although most off the shelf cameras can be usedthe Intel PC Camera Pro was found suitable in thatit has very good image quality and can focus down
to 25mm. The device has a standard USBinterface. Good results have also been obtainedfrom other cameras including models from
Logitech and Philips. All the cameras have thebenefit of being low cost consumer items availablefrom around 600 rand. The cameras simply plug
into a standard USB port and require no additionalhardware or complex configuration.
The software has the provision to calibrateresistance thermometers and a maximum of 16can be connected to external selector switch or
multiplexers, Isotech model 954.
Using an indicator that can read both
thermocouples and RTDs increases the numbersensor types that can be calibrated and there is acompatible thermocouple selector switch, model
958.
Other models of temperature baths can be used
so that the system can cover temperature rangesfrom -75C to 1300C.
Software
The I-cal software sets the calibration bath to thefirst temperature and monitors the temperature
controller, when the set temperature and theindicated temperature are within a user definableband for a chosen time period the software
monitors the standard. The user can specify anumber of samples and a variation, when thestandard has met its criteria the data is logged
from the instrument and an image is grabbed fromthe camera. This image is captioned with theactual temperature as measured by the reference
thermometer and the image is saved to the PCwith the file name consisting of the date and timeof the capture.
The provision of the user being able to specify thestability criteria allows the use of differentcalibration baths and temperature ranges. For
example the criteria for liquid bath would besmaller than the appropriate setting for a hightemperature thermocouple calibration furnace.
The self documenting images can be verifiedagainst the computer log file which also contains
the value read from the TTI-6 along with the dateand time.
The I-cal software allows for a number of imagesto be captured at each calibration point. These
images can be examined to see if the display wasstable during the unattended calibration.
I-cal also has provision to upload images to a local
or remote server which allows the system to bemonitored over a network or the internet. This hasproven useful for monitoring equipment away from
the laboratory and opens the possibility of futureremote calibration.
-
8/12/2019 Auto Image
3/5
Isothermal Technology Ltd http://www.isotech.co.uk
Determining the Uncertainty
The measurement system, TTI-6 and Resistance
thermometer have UKAS accredited calibrationand the uncertainty can be taken from thecalibration certificate.
The thermometer under test is compared to theTTI-6 and the Zeroth Law of Thermodynamics is
relied upon in the assumption that the testthermometer and the standard thermometer are atthe same temperature.
The Zeroth Law of Thermodynamics states , "Iftwo systems are in thermal equilibrium, each
having the same temperature as a third system,the two systems have the same temperature aseach other".
For calibrating a thermometer in the Dry Block, thelaw could be written as, If two thermometers are
in thermal equilibrium, each having the sametemperature as the Dry Block, the twothermometers have the same temperature as each
other.
Each Calibration Bath will have special
characteristics of temperature distribution in theblock of the calibrator. If the Dry Block has acalibration certificate that specifies an uncertainty
including the temperature distribution in thecalibration area this value can be used.
The combined uncertainty of the system can thenbe established by combing the uncertainties of thestandard and Dry Block using the RSS method.
uuu cntdbstdUct222
++= (1)
Where
Uct= Combined System Uncertainty
ustd= Uncertainty of the Standard, TTI-6 and
probe
udb= Uncertainty of the Dry Blockucnt=Uncertainty of the connections
The uncertainty associated to the calibration
remains controversial. Simplistically in the UK aUKAS accredited laboratory would include theuncertainty of the unit under test with the
calibration uncertainty and specify this on thecalibration certificate. Other laboratories may omitthe uncertainty of the unit under test and state only
their ability to create a know isothermal condition.
How to Estimate the Dry Block Uncertainty
If the Dry Block uncertainty is not known it must becalculated. The European co-operation for
Accreditation (EA) has published Guidelines on
the Calibration ofTemperature BlockCalibrators [2].
In general the largestuncertainty will be the
vertical temperaturedistribution or Axialtemperature homogeneity
along the boring in themeasurement Zone. This isusually the largest
uncertainty source. Figure 1illustrates this parameter.
It will be most significant whenthe standard and the unit under
test have different properties.
Figure 2 shows athermocouple with shortsensing length being
compared to a resistancethermometer which has aninternal sensing element of
15mm. The thermocouple issensitive to changes intemperature from its
junction whilst theresistance thermometer willto a large extent integrate
the temperature over itslength.
The measurement zoneshould be specified by themanufacturer and should
not be less than 40mm.
Axial temperature
homogeneity along theboring in the measurementZone. This is usually the
largest uncertainty source.
Temperature differences
between the boringsFor most type and brands of Dry Block thisparameter is smaller than the axial homogeneity,
see figure 3. This parameter can be measuredrelatively easily. A good method is to use two
Figure 1
Figure 2
-
8/12/2019 Auto Image
4/5
Isothermal Technology Ltd http://www.isotech.co.uk
probes moving them between the pockets ofinterest and calculating the difference,
Dt, using: -
Dt = [(TA1 TA2) + (TB1 TB2) (2)
TA1 is Thermometer A in
Pocket 1TA2 is Thermometer A inPocket 2
TB1 is Thermometer B inPocket 1TB2 is Thermometer B in
Pocket 2
Influence upon the
temperature in themeasurement zone due todifferent Loading
This value is determined bytaking measurements with different block loadings.Using an external reference thermometer (not the
Dry Blocks internal control sensor) reduces thiseffect.
Stability with time The variation with time of theDry Block will introduce a further uncertainty. Theeffect will be most significant when the standard
thermometer and the unit under test have differenttime constants. Stabilities of +/-0.02C over 30minutes are readily achieved with modern control
techniques.
Temperature deviation due to heat conduction
There will be a flow of heat along the thermometerstem. As the difference between the environmenttemperature and the Dry Block temperature
increases the error will proportionally increase. Tominimize this effect, Stem Conduction orImmersion Error the thermometers need to be
adequately immersed. For probes of 6mm andless the EA Guidelines state a minimumimmersion depth of the thermometer to be
calibrated is at least equal to 15 times the outside
diameter of the thermometer to be calibrated.
Other Uncertainty Sources
Uncertainty of the Standard Thermometer
Standard Thermometer including measurementwith standard thermometer this can usually betaken from the calibration certificate.
Hysteresis The temperatures indicated may showa deviation due to hysteresis in cycles of
increasing and decreasing temperatures. This canbe ascertained by taking measurements as the
temperature decreases after excursion to themaximum temperature.
Resolution The uncertainty due to the resolutionof the device under test needs to be taken intoaccount, the uncertainty contribution form the
standard being included in the standards ownuncertainty budget.
The uncertainty, Uct, is calculated by:
Uct = std+ti+tR+tH+ tB+t+tV (3)
where:
Future Developments
It has been suggested that Optical CharacterRecognition and the support of multiple cameraswould enhance the system. These could be
implemented but it would move the project awayfrom the starting point of a low cost generalpurpose system.
Conclusion
A simple low cost system has been built to capturedata from indicators without a PC interface. Low
std - temperature of the referencethermometer derived from theresistance measurement includingtemperature correction and an
allowance for drift since its lastcalibration
ti - temperature correction due to limited
tR - Temperature difference betweenborings
tH - temperature correction due tohysteresis in the increasing and
decreasing branches of themeasuring cycle;
tB - temperature correction due to axial
inhomogeneity of temperature in theborings
tL - temperature correction due todifferences in the loading of the blockwith thermometers to be calibrated
tV - temperature variations during thetime of measurement.
Figure 3
-
8/12/2019 Auto Image
5/5
Isothermal Technology Ltd http://www.isotech.co.uk
cost consumer digital cameras can be used toperform automatic temperature calibration.
References
[1] C. Dawn Vaughn and Gregory F. StrouseTHE NIST INDUSTRIAL THERMOMETER
CALIBRATION LABORATORY National Instituteof Standards and Technology, TEMPMEKO 01
[2] European co-operation for Accreditation (EA)EA-10/13 Guidelines on the Calibration of
Temperature Block Calibrators, FEB 2000