·Table of Contents
·Workshop - Visual testing
Verification of the resolution capability for equipment used in Visual TestingDe Petris C.
I.S.P.E.S.L. - Department of Safety Technology - Italy
Rolls Royce Plc. Naval Marine, Operations Division - United Kingdom
|1. the visible minimum can be defined as the capability to detect only the presence of objects in the field of view but not necessarily recognising or classifying them. It could be also called detection capability;
||2. the separable minimum can be defined as the capability to resolve single separated parts of a complex image. It could be also called resolution capability;
|3. the visual acuity by Vernier can be defined as the capability to perceive the smallest variation between two objects in the spatial ratio. It could be also called perception capability of variation;||4. the readable minimum can be defined as the capability to recognise the meaning of a group of marks like a letter or a number. It could be also called recognition capability.
Among the 4 perceptions mentioned above, it is evident in importance to assess at least the resolution capability (item 2) since it is certainly the most significant. To investigate this aspect in more detail, several resolution targets have been developed and a simple experimental apparatus has been arranged which will be outlined further below.
|Picture||Controls the digitization of video: |
live video, image capture, image averaging, input channel selection, setting input look-up tables, setting gain, offset, and reference values, linear histogram equalization
|Display||Controls the display of images: |
buffer selection, setting output (color) look-up table, zooming and panning, copying and pasting images
|Profile||Graphs the grey values along a line segment|
|Histogram||Graphs the distribution of grey values in a region of interest (ROI) and calculates intensity statistics|
|Location||Displays the position and grey value of points|
|Measure||Displays the distance between two points, between a series of points, along a freehand line, or along an edge in an image and displays angle measurements|
|Filter||Applies convolution and morphological filters to an ROI|
|Arithmetic||Performs arithmetic and logical operations on an ROI, and copies an ROI to another buffer|
|Geometry||Rotates, flips, or scales an ROI|
|Draw||Draws text and simple graphics on an image and allows greyscale image editing|
|Log||Collects and displays the data logged from other tool windows|
|Calibration||Calibrates pixel coordinates to known world coordinate system|
|Frequency||Computes forward and inverse FFTs, and performs frequency editing on an ROI|
|Particle||Counts, measures, and classifies particles in an ROI|
The experimental configuration includes other devices as shown in fig. 1:
The experimental tests have been carried out using a "black room" consisting of six square wooden panels forming a square box (fig. 2). One of the panels can rotate on a clasp to allow an easy internal access and to fix a resolution target by an adhesive tape. On the external side of this panel a little permanent magnet has been fitted in order to allow the positioning of a light metallic demonstration testpiece. The camera head was installed at the end of a ruled sliding guide located at the centre of the opposite panel and moving perpendicularly to the resolution target surface to set up the distance of the image sensor to it. Four lamps, whose light intensity can be varied by a rheostat, were placed inside the box at the four corners of the same panel supporting the sliding guide. Close to the target or demonstration testpiece, a lux meter photocell provided a measurement of the illumination during tests.
|Fig 1:||Fig 2:|
During tests, the following parameters have been considered:
Before each experimental test, an accurate set up of the camera focus has been ensured.
The most significant symbol-letters and marks used for the development of the resolution targets are the following:
|Fig 3a:||Fig 3b:||Fig 3c:|
The first resolution target (fig. 4a) can be properly defined as a line chart where both line and gap thickness t increases from 0.1 to 1 mm with a step of 0.1 mm. Differently, the second (fig. 4b, with Landolt ring optotype) and third (fig. 4c, with Albini optotype) resolution targets copy the same frame of the previous one but with a different progression in size and a sequential arbitrary orientation.
|Fig 4a:||Fig 4b:||Fig 4c:|
As it's known, under the clinic point of view the visual acuity grade 1 shell be represented by a Landolt ring whose outer diameter subtends an angle of 5' and whose width, as well as the gap in its continuity, subtends an angle of 1' at the designated viewing distance [Ref. 2, 3]. This means that an observer placed 5 m distance (see fig. 5) from the panel on which several symbols are drawn and under daylight conditions should be able to discriminate the position of a gap 1.46 mm high on an optotype 7.3 mm high (note: the ratio between the two heights is 1/5). So, assuming 1 minute of arc as a reference condition for the gap and the extent of a possible resolution target format with the previously mentioned characteristics in terms of backgrounds, an increasing progression of sizes has been proposed and listed in Table 2. The correspondent distances proportionally achieved to the smallest size evaluated at 600 mm have also been indicated.
Fig 5: Standard Condition|
Resolution Step 1.46mm
|IDENTIFICATION NUMBER||DISTANCE [mm]||H|
RESOLUTION STEP [mm]
|Table 2: Refer to the resolution targets shown in fig. 4b and 4c|
It shall be stressed that this choice has been absolutely arbitrary and suggested by simple reasons of opportunity.
The results of tests have been obtained under the following conditions for the test parameters:
The fig. 7 shows a typical distribution of grey levels along a line segment transversally crossing a sequence of line bands as well as the digital image acquiring and processing equipment provided. In particular, the curves have been achieved at different contrast backgrounds (from 0 to 50% of black) and under the same conditions for illumination (300 Lux). The values of the grey level are within a range 0 - 255 expressed in digital units depending on the performance of the digital image acquiring and processing equipment. It can be noticed that for this monochromatic grey scale the level 0 corresponds to the fully black and 255 to the fully white. It must be stressed that no correspondence exists between the grey scale used to create the backgrounds of the monochromatic line chart and the grey scale pertaining to the digital image acquiring and processing equipment. That explains why the part of the curves shown in fig. 7 relative to the monochromatic analysis on the backgrounds has got grey levels expressed in digital units higher than those corresponding to the black bands of the line chart.
|Fig 7: Illumination 300 Lux Target to image sensor distance 600 mm|
It is interesting to note that the measured grey level pertaining to the white background (Contrast 100) is constant at 255 digital units denoting a condition of saturation, while reducing the contrast it slowly begins to oscillate around a decreasing average value. The explanation of this circumstance shall obviously be searched in the combination of the technical characteristics of the equipment used to perform the visual test in relation to the operating conditions. Even if it could be interesting to probe into this aspect of the problem, it can not be examined here because it is not to do with the subject of the present paper. Anyway, the normative explicitly requires a verification performed as follows: "Equipment and the conditions shall be arranged as per the actual test with the whole system" and "The verification test parameters and results of the verification shall be included in the post test documentation".
Carrying on the analysis of the curve shown in fig. 7, it is easy to recognise the zones of it where the line segment transversally crosses the bands containing the black lines. They are characterised by a sequence of peaks more accentuated in terms of grey level amplitude. It is also confirmed that an expected attenuation of the amplitude for the thinner lines up to a band level where the system does not resolve a single line any more and compacts them in an associated meaningless signal. This exhibits an evident limit of the system for those specific test conditions. In fact, not all lines relative to the bands having lower thickness are distinctly recognisable and so an acceptance criterion shall be stated for the determination of the limits within which the equipment could be deemed suitable to perform a designated visual task. This acceptance criterion is obviously more or less restrictive depending on the subjective judgement of the operator or defined in an agreed procedure. The criterion adopted here is to accept a resolution level corresponding to the band whose thinner thickness of its lines and gaps provide all four correspondent distinct relative minima. The activity so arranged has been carried out under further different illuminations but holding the other tests parameters constant. Fig. 8 and 9 respectively show the experimental results obtained for 600 Lux and 900 Lux. It makes evident the effect of glare on the resolution capability when the illumination increases as well as the opposing role played by the contrast.
|Fig 8: Illumination 600 Lux Target to image sensor distance 600 mm|
|Fig 9: Illumination 900 Lux Target to image sensor distance 600 mm|
At the end of the extended experimental activity, a 3 D map of the allowed conditions of the investigated parameters for the equipment used to perform tests can be developed. Fig. 10 shows a synthesis of the actual empirical results where the possible acceptable conditions for tests parameters can be found only above the lines connecting the points achieved by the application of the criterion previously mentioned.
A similar approach can be followed when another resolution target is used. When alternative targets shown in fig. 4b or 4c are selected, the same criterion previously presented can be adopted even if a threshold of contrast in terms of grey level between the relative minimum and the average values pertaining to the background has to be assumed or fixed.
It should be empathised the particular simplicity to arrange the experimental system and the ease to make the resolution targets shown here. It should again be stressed that the size of the symbols or optotype used to create the resolution targets have been assumed here arbitrarily and can be of course be changed depending on the task to be performed. Notwithstanding, the possibility to fit a grey scale to the backgrounds which can be a way to analyse the effect of the contrast.
An experimental activity planned as described can be useful to determine the effective limits of the system used to perform visual testing when the acceptance criteria have been defined.
|© AIPnD , created by NDT.net|||Home| |Top||