WO2016181099A1

WO2016181099A1 - Night vision device testing

Info

Publication number: WO2016181099A1
Application number: PCT/GB2016/050754
Authority: WO
Inventors: Laurence Durnell
Original assignee: Fenn Night Vision Limited
Priority date: 2015-05-14
Filing date: 2016-03-18
Publication date: 2016-11-17
Also published as: GB2532651A; GB2532651B; GB201602864D0; GB201508238D0

Abstract

A system for testing a night vision device (104) comprises a camera (106) arranged to detect an image produced by the night vision device (104) and to output image data encoding the image, and a tablet or other processing device arranged to analyse the image data to determine whether the image meets at least one criterion, and to generate an output indicative of whether the at least one criterion is met. The criterion may relate to the presence of blemishes or hexagonal patterns in the image, or the resolution of, or distortion caused by, the night vision device (104).

Description

Night Vision Device Testing

Field of the Invention

The present invention relates to night vision devices (NVDs), such as night vision goggles (NVGs), and in particular to the testing of NVDs.

Background to the Invention

Night vision devices, including binoculars and monoculars as well as goggles, are much used in military and other environments to provide improved vision in low light environments. They use well known image intensification methods to produce, from a low intensity image which the human eye cannot easily see, a higher intensity image which it can.

NVDs do not produce perfect images, and various methods are known for testing the quality of the images that they do produce . Common problems with the images include poor resolution, the presence of blemishes in the images, i.e. small features in the intensified image which do not correspond to anything in the original image, and also the presence of a hexagonal 'honeycomb' or 'chicken wire' pattern in the intensified image which arises due to the structure of the micro-channel plate used for image intensification and becomes visible under some conditions.

It is known to check for these features manually by looking through the NVD at either an array of blocks or bars of various sizes to identify the smallest which can be seen thereby to determine the resolution, and by looking for and counting blemishes, and by checking for visible 'chicken wire' patterns. However, these tests are inevitably somewhat subjective. There are also other problems with the NVD which are hard, or in some cases impossible, to detect visually, such as gross distortion, discontinuities, flicker, and misalignment and variation in magnification between the two sides of a binocular NVD.

Summary of the Invention

The present invention provides a system for testing a night vision device, the system comprising detection means arranged to detect an image produced by the night vision device and to output image data encoding the image, and processing means arranged to analyse the image data to determine whether the image meets at least one criterion, and to generate an output indicative of whether that at least one criterion is met.

The processing means may be arranged to identify the presence of at least one feature in the image. The at least one criterion may be the presence of a feature, or the presence of at least a predetermined number of features.

The system may further comprise a display screen and may be arranged to provide an indication on the display screen of whether the at least one feature is detected.

The system may be arranged to display the image on the display screen, and may be arranged to highlight the feature on the image .

The system may further comprise image generating means arranged to generate an input image so that the image produced by the night vision device is an output image reproducing the input image.

The input image may be a plain image, for example a plain white or other pale colour image. The processing means may be arranged to identify features in the output image which are not present in the input image.

The processing means may be arranged to identify features meeting at least one feature criterion, and to determine the number of such features in at least an area of the image. The features may be parts of a hexagonal 'chicken wire' pattern, in which case the number of them that are identified will indicate the extent or strength of the pattern. Alternatively they may be blemishes, which can be identified by various criteria. The processing means may be arranged to control the display so as to highlight each of the features that meet the at least one criterion. The input image may comprise a plurality of features of the same shape but of different sizes, or different orientations, or both, and the processing means may be arranged to determine which of the corresponding features in the output image meet the at least one criterion. For example a set of criteria may be used to define the feature, which may itself comprise a number of sub-features, such as a number of bars or lines. The size of the smallest feature that can be identified will be an indication of the resolution of the NVD.

The processing means may be arranged to measure a parameter of the output image and to calculate from the parameter the gain of the device .

The processing means may be arranged to measure a parameter of the image or the output image, and to calculate from the parameter the level of gross distortion produced by the device.

The processing means may be arranged to detect discontinuities produced by the device in the image or the output image .

The processing means may be arranged to measure the magnification of the device.

The processing means may be arranged to measure alignment between two sides of a binocular night vision device.

The processing means may be arranged to measure flicker in the image or the output image.

The processing means may be arranged to measure signal to noise ratio in the image or the output image. The present invention further provides a system for aiding a user in focusing an optical device, the system being arranged to acquire a sequence of images output by the device, calculate a focus metric for each of the images, and generate a user- perceptible output indicative of the focus metric. The output may be arranged to vary in real time during adjustment of the focus of the device .

The output may be a plot of the metric against time, for example a line plot or a bar graph. The system may be arranged to identify a maximum value of the metric as the device passes through a point of maximum focus, and then to detect when the value of the metric returns to that maximum value and in response to generate a further user perceptible output. The system may further comprise any one or more features, in any combination, of the preferred embodiments which are shown in the accompanying drawings, as will now be described.

Brief Description of the Drawings

Figure 1 is a schematic diagram of a system according to an embodiment of the invention;

Figure 2 shows an example of a test image as displayed on the screen of the system of Figure 1 ;

Figure 3 shows an example of a further test image as displayed on the screen of the system of Figure 1 ;

Figure 4 is a chart showing the steps of a test procedure performed by the system of Figure 1 ; Figure 5 is a display produced by the system of Figure 1 during the test procedure of Figure 4;

Figure 5a shows the variation in intensity over one of the bars of the image of Figure 2, as measured in the procedure of Figure 4;

Figures 6, 7, 8, 9 and 10 are charts showing the steps of further test procedures performed by the system of Figure 1 ;

Figure 11 shows a test image used in the test procedure of Figure 10;

Figure 12 shows an output image from the device, analysed in the test procedure of Figure 10;

Figure 13 is a chart showing the steps of a further test procedure performed by the system of Figure 1 ; Figure 14 shows an output image from the device, analysed in the test procedure of Figure 13 ; Figures 15, 16, 17, and 18 are charts showing the steps of further test procedures performed by the system of Figure 1 ;

Figure 19 is a schematic top view of a system according to a further embodiment of the invention; and

Figure 20 is a schematic section through the system of Figure 19. Description of the Preferred Embodiments

Referring to Figure 1 , the test system comprises an image generation device 100 which includes a display screen and is arranged to display one or more images, and may comprise a mounting 102 arranged to support an NVD 104 so that the image displayed on the image generation device 100 is in the field of view of the NVD 104, and forms an input image for the NVD 104. The image generation device may be a dedicated device for testing of NVDs, and in this embodiment comprises a Hoffman 126A test set. The system further comprises an image detection device, such as a digital camera 106. This may be, for example, an IDS UI- 1490SE, or other suitable machine vision camera. The system may further comprise a further mounting device or connector 108 which is arranged to connect the camera 106 against the eyepiece 1 10 of the NVD so as to detect and record the output image generated by the NVD 104, and to output image data encoding the image. For a binocular device, two such cameras may be provided, one mounted on each side of the NVD. A processing system or computer device, in this example a tablet 1 12, may be arranged to receive mage data from the camera or cameras 106, and may include a processor 1 14 and memory 1 16 and a display screen 1 18. It may also be arranged to control various aspects of the operation of the camera or cameras 106, such as the frame rate and exposure time of the camera. It may also be arranged to control which of the detectors from the full array of the camera are used for each frame, so that, for example, each frame can include the maximum number of pixels corresponding to the total number of detectors in the camera, or a smaller number of pixels using only some of the detector array. This can allow smaller images to be acquired with a higher frame rate than is possible for the maximum image size.

The test set 100 may be purely for monocular testing, and may generate only a single image at one time. Alternatively it may be for binocular testing, and may therefore generate two images simultaneously, one for each side of a binocular NVD. Operation will be described below for a single monocular system, but for each test it will be appreciated that both sides of a binocular system may be tested in the same way, either simultaneously, or sequentially but without moving the NVD. A sliding mounting for a binocular NVD may also be incorporated in the system so that the two sides can be tested sequentially using the same input image .

The test set 100 may be arranged to perform a number of functions, but for the purposes of the present invention, the relevant functions may include the display of various types of input image, one of which may be a plain image, for example a white image with a number of rings defining different regions of the image, against which blemishes and chicken wire artefacts can easily be detected, one of which may be a resolution checking image which may contain features of a range of sizes, for example as described in more detail below, and one of which may be a rectangular or square grid.

The tablet 1 12 may be arranged to run a software program or application which is arranged to analyse the image data from the camera 106 to test the NVD 106 to determine whether the data, and therefore the output image, meets at least one test criterion. The program may provide a number of functions which can be selected by a user, for example using icons 120 displayed on the screen 1 18 of the tablet. The embodiment shown provides all of the functions which will be described below, but other embodiments may offer only some of these, and may offer others. Referring to Figure 2, as mentioned above, one of the images that the test set 100 may be arranged to display is a resolution testing image to test the resolution of the NVD . The test set 100 may generate an image which is shown in Figure 2 which comprises a number of sets of bars, for example three bars 200. Each bar 200 is a solid rectangle or other suitable shape, of black or another suitable colour, and each set 200a, 200b may comprise three identical bars arranged parallel to each other with a small gap between them so that the whole set covers an approximately square area. The sets 200a, 200b vary in size, but for each size there may be a pair of two sets 200a, 200b one with the bars arranged vertically and one with the bars arranged horizontally. The pairs may be arranged in groups of four pairs, and in Figure 2 two groups 202, 204 are shown, with the largest sets of bars in one group 204 being smaller than the smallest sets in the other group 202.

When the test program is running on the tablet 1 12, one of the icons 120 can be selected by a user to test the resolution of the NVD, and this is selected when the test set 100 is displaying the resolution testing image shown in Figure 2. The program may then be arranged, using image processing techniques such as edge detection and shape recognition, to detect as many sets 200a, 200b of bars 200 as it can, or to measure the resolution of the bars to determine which of them are imaged with at least a threshold level of resolution. For each set that it detects, or detects with sufficient resolution, it may be arranged to identify that set, for example by highlighting it on the screen, in the example shown this is by displaying a box 206 around it. Obviously other methods of highlighting the identified sets can be used, for example by changing their colour or changing the background around them. Once the program has identified as many sets of bars as possible, the results of the test can be displayed in a number of ways. For example the highlighted sets 200a, 200b can simply be displayed so that a user can determine which ones have been identified, and therefore determine the resolution of the NVD in a similar way to when the user is looking at the sets of bars and determining themselves which ones can be seen and which not. Alternatively, the program can be arranged to display details of the smallest set or sets, or the smallest group and all of the identified sets within that group, for example just indicating them by number. In the example shown, a display area 210 on the screen indicates that sets a, b and c from Group 2 are the smallest sets that have been identified. This has the advantage over known methods that the user does not have to decide which of the sets is clear enough to be distinguished and which not. An example of how this test can be performed is described in more detail below with reference to Figure 6.

Referring to Figure 3 , in a further test, the test kit 100 is arranged to display the plain white test image with circular regions defined on it as described above. This means that any blemishes or artefacts that the NVD superimposes on the input image can be easily identified. A completely plain image could be used instead. The tablet is then arranged to use image processing techniques to identify two specific features in the output image: hexagonal patterns, commonly referred to as 'chicken wire' or 'honeycomb' patterns, and blemishes. The chicken wire pattern 300 is very distinctive and the program is arranged to detect individual lines 302 including their position and orientation, and then to analyse them to determine whether they are part of 'chicken wire ' patterning. In this case the tablet is arranged to do this by detecting the presence of, for example, groups of three lines 304a, 304b, 304c meeting at a point at angles of 120° to each other. Once a sufficient number of these has been detected, the program is arranged to determine that the 'chicken wire' pattern has been detected. The strength of the pattern can be determined by, for example, the number of instances of the required combination of lines being detected in the whole image, and the thickness or darkness of the lines compared to the background. The program can then compare the detected pattern against one or more threshold criteria and give a pass or fail indication. Alternatively, or in addition, it can be arranged to grade the severity of the chicken wire pattern, for example on a scale of 1 to 6. An example of how this can be done is described in more detail below with reference to Figure 7. The results can then be indicated to the user, for example in a results area 306 on the screen 1 18.

In order to identify, categorize and count the blemishes 400, the program is arranged to identify, again using edge detection, shape recognition and other suitable algorithms, any features (other than the hexagonal pattern) which are present in the image, and to determine their size. Once the number and size of the blemishes has been determined, this can be displayed to the user, either by simply highlighting them so that the user can count them, or by giving a direct indication of the number in a dedicated area of the screen 1 18. Obviously a variety of criteria can be used to determine what should be counted as a blemish, for example it might include only shapes over a particular area and/or length. Also each detected blemish can either just be identified as a blemish, or it may be categorized, for example by size and/or shape, and the number of each category of blemish indicated to the user, either just as a number for each category indicated in the results area 306 on the screen, or by highlighting different categories in different colours. An example of how the test for hexagonal patterning is performed is described below with reference to Figure 8.

While it is possible for the chicken wire detection and blemish counting to be performed for the whole of the image, it can be more useful for a separate analysis to be carried out for different parts of the image . For example, the image may be divided into a plurality of different areas 402, which may be indicated on the image, for example by concentric rings 403 included in the white test image as shown in Figure 3, or rectangles or squares or segments. The determination of the presence or absence and categorization of chicken wire patterning, and the number and nature of blemishes, is then performed for each area, and the results indicated separately for each area in the results area 306 of the screen. These lines are used if the test is done by eye, but as will be described in more detail below, they can be used for other automated tests performed by the system.

Referring back to Figure 1 , the eyepiece 1 10 of the NVD is typically manually adjustable to adjust the focus of the output image produced by the NVD. The system may be arranged to perform a focus assist function which aids a user in focusing the NVD so that other tests can be performed on it. Referring to Figures 4 and 5, the program is arranged to perform a calculation of focus on an output image, for example generated using the resolution test image of Figure 2, and to repeat this for subsequent images over a test period while the user adjusts the focus, and indicate the result of the focus calculation, and how it varies over time, to the user on the screen 1 18. As shown in Figure 5, the result may be displayed in real time, for example on a plot of focus against time, which may have time on a horizontal axis and focus, as determined by a focus metric, on the vertical axis. The user can then watch the plot as the focus is adjusted to determine when the optimum focus is reached.

As shown in Figure 4 the program may be arranged to acquire an initial set-up image data set for an image of the test pattern with a reference exposure, and from that data set to calculate the best exposure settings. The program is then arranged to communicate those optimum settings to the camera which is then arranged to use them for the rest of the test. The program then selects a test frame rate which is as fast as possible, whilst ensuring that the contrast is sufficient for the focus of each image to be determined. This may comprise testing the contrast between the lightest and darkest parts of the set-up image and ensuring that it is above a predetermined threshold, which is sufficiently great for the focus measurement to work. Generally the greater the frame rate, the lower the contrast will be. Once the optimum test frame rate has been set, the program is arranged to acquire a series of test images at the test frame rate, and for each test image, to calculate the value of a focus metric and display the value on the screen as shown in Figure 5. The value of the metric for each frame may be displayed as soon as it is calculated so as to generate a real-time plot of the focus. As the value of the metric changes over time, the plot will rise and fall, and the user can monitor this while adjusting the focus to determine where the best focus is.

For example as shown in Figure 5, as the eyepiece 1 10 is adjusted the focus might increase towards the maximum, and then decline again which will be seen as a rising then falling curve 500. From this the user can see approximately where the maximum value 502 of the focus metric occurs, and move the eyepiece back towards the maximum focus. Then as the focus passes through the maximum again at point 504 this can be seen and the focus adjusted back again to the maximum. Obviously in practice the user may move the focus through the optimum any number times, but with practice the optimum focus can be found very quickly. The program may also monitor the variation in focus metric over time and may detect the maximum value of the focus metric, for example by identifying the top 502 of the curve as the focus moves through that point. In may then be arranged to store the maximum value, and generate an output to the user when that maximum value is again reached, for example as an audible feedback, or visual feedback on the screen 1 18. This can aid the user in determining when the optimum focus has been reached.

The focus metric may be calculated in a number of ways, but may for example comprise plotting the intensity or grey level of the output image of one of the bars 200a in the test image of Figure 2, which will generally have the shape shown in Figure 5a with a high level 5 10 outside the bar and a low level 5 12 within the bar. If the focus is good the plot will be a sharp, deep, rectangular dip, whereas if the focus is poor, the dip will be rounder and more shallow. Therefore the focus metric may be the difference in intensity between the two levels 5 10, 5 12, or may include a measure of the sharpness of the intensity dip. Once the focus of the NVD is optimized, then the system can perform a number of other tests.

Referring to Figure 6, the automated resolution test, described in general terms above, will now be described in more detail. Firstly a sequence of images of the test image of Figure 2 is acquired. The program may then be arranged to average the sequence of images, for example by calculating an average over all the images for the value of each pixel. It may then be arranged to normalise the average image to correct for any variation in brightness over the area of the image, which is typically brighter in the centre than towards the edges, thereby producing a sample image. It may then be arranged to locate one or more large features in the image, such as the largest group of bars, and determine the exact position within the image of those features. Using stored data, for example a stored copy of the test image or stored coordinates of all the features of the test image, the program may calculate the position, scale and location of the image, i.e. the exact position, scale and location within the area imaged by the camera, of all of the features of the test image. From this it may determine where within the sample image all of the features of the test image should be. The program may then test the contrast of a number of features in the sample image, and from the results provide an indication of the resolution of the NVD . The program may be arranged to measure the contrast of one of the bars 200 of the test image as it appears in the sample image by determining the brightness at a point in the centre of the bar, and the brightness of a point well outside the bar. This is similar to the focus test described above. As described above with reference to Figure 5a, if the resolution is good, the contrast will be high as the centre of the bar will be black and the area just outside the bar will be white . However, if the bar is not well resolved in the sample image, the centre of the bar may not be completely black as the image will be somewhat blurred. Therefore the comparison between the two brightness levels gives a measure of the contrast for that bar. Generally the centre of larger bars will be darker than the centre of smaller bars, so the program may be arranged to set a threshold contrast, and then measure the contrast for each of the bars, and determine the smallest bars for which the contrast is above the threshold level. This result can then be indicated to the user as described above.

Referring to Figure 7, the honeycomb pattern test may be performed as follows. The tablet may first acquire a single sample image of the white test image. If the test image includes circles or other lines defining areas of the test image, then the program may be arranged to locate those and remove them to form a plain sample image. The program may then be arranged to calculate the extent of the honeycomb pattern in the plain sample image, and to allocate the extent of the pattern to one of a plurality of categories, for example on a scale of 1 to 10. This may then be repeated for a number of separate test images. The scale value of the honeycomb pattern for the images may then be averaged, and the program may then determine whether the NVG passes or fails this test based on the average value . It is possible to average the sample images and then calculate the extent of the honeycomb pattern on the averaged image . However this is less effective as the averaging tends to remove the honeycomb pattern. The results can be displayed to the user on the screen 1 18 either as a simple pass/fail indication, or as graphical slider on a scale so that the strength of the honeycomb pattern is indicated, either on a continuous scale or as one of a number of finite steps or categories, so that a user can tell whether the test is passed or failed, and how marginal the result is.

Referring to Figure 8, an example of how the blemish test can be performed by the program will now be described. The program may first be arranged to acquire a sequence of sample images of the plain test image, and then to average those images to form an averaged sample image, which may then be normalised as for the resolution test described above. The images may be acquired using high 'gamma' which allows features in the darker parts of the image to be relatively easily identified. The average sample image may then be converted to a binary image in which each pixel has one of only two values, which can be represented as black and white . This will generally result in a sample image which is mostly white with black blemishes of various sizes and shapes in it. All areas of black may then be identified by the program, which may be arranged to categorize them by size and shape, in a 'blob labelling' step. Any circles or other lines intentionally present in the plain test image may then be identified by their size and shape, and excluded from the calculation, or they may be left in, for example if the blemishes are to be identified to a user for counting. Once all the blemishes have been categorized, they may be identified by category, for example by defining a plurality of size categories, assigning a colour to each category, and highlighting all the blemishes in each category in the colour associated with that category. If the areas of the image are also displayed this enables a user to count the number of blemishes in each category in each area. Alternatively, or in addition, the program may be arranged to determine the number of blemishes, in total or in each category, in each area, and indicate this numerically. Alternatively, or in addition, the program may be arranged to compare the number of blemishes in each area with a reference threshold number for that area, and if one or more of the thresholds is exceeded to determine that the NVD fails the test, but if none of the thresholds is exceeded, to determine that the NVD has passed the test. The pass/fail test result may be indicated to the user on the screen in any suitable manner.

Where the NVD is a goggle or other binocular device having two imaging systems, one on each side, one for each eye of the user, the program may also be arranged to perform a gain test arranged to compare the gain of the two imaging systems. The gain referred to here is the increase in brightness of the output image from the NVD compared to the input image . Clearly it is beneficial for the gain of the two sides to be equal. For this comparison to be accurate, the same input image may be used for both sides of the NVD, and the same camera may be used to capture the output image from both sides. This may be achieved, for example, using a sliding mounting for the NVD that enables the NVD to be moved between two positions, in one of which one side of the NVD is aligned with the input image so that it will produce an output image for capture by the camera, and in the other of which the other side of the NVD will be so aligned. To perform the test, the system is arranged to acquire an image with each side of the NVD, and then from those images to calculate a measure of the gain of each side of the NVD. The gains can then be compared, either by the system or by a user, to determine whether they are close enough to each other. Referring to Figure 9, to perform the gain test the program may be arranged to acquire a sequence of output images for one side of the NVG, for example the left side, which are all from the same input image which may for example be the resolution test image. It may then be arranged to calculate an average image from the sequence of images . It may then be arranged to calculate an average grey level for an area of the average image, for example a small central area of the average image. It may then be arranged to convert the average grey level value to an average luminance value . These steps may then be repeated for the other side of the NVG to determine an average luminance value for that side . It may then be arranged to compare the two luminance values, for example by calculating a difference between the two values as a percentage of one of them or of the average of the two. This percentage may then be compared with a threshold percentage, and if it is below the threshold to determine that the two values are similar enough to pass the test, but if it is above the threshold to determine that they are too different in which case the test is failed. This pass/fail determination may be indicated to the user by any suitable indicator, for example on the screen 1 18, or the percentage value itself may be indicated. Referring to Figures 10 to 12 the program may also be arranged to perform a gross distortion test to measure the degree of gross distortion of the image. Gross distortion occurs in the inverter or 'twister' which is a twisted fibre optic bundle arranged to correct for inversion of the image caused by the objective lens of the NVD . Slight imperfection of the twister results in distortion of the final image which converts straight lines to a generally S-shaped line . To perform this test the test kit 100 may be arranged to generate an input image which includes a number of straight lines 1 10, for example in the form of a square grid 1 12 as shown in Figure 1 1. The test program may be arranged to acquire a sequence of output images from the NVD using the camera. It may then be arranged to average those images and normalise them as described above . It may then be arranged to locate the grid lines 122 in the output image. As these may be distorted by gross distortion, as shown in an exaggerated form in Figure 12, the search may be arranged to find not just straight lines, but any lines which are straight or curved. Once the grid lines 122 have been located in the output image, the program may be arranged to calculate the distortion at a number of points along the lines, and convert the lines to data points on a distortion graph. To calculate the distortion, the program may be arranged to calculate the horizontal offset dx at points along each of the vertical lines 124 in the output image from the original position of the corresponding line 124a in the input image, and to calculate similar offset in the vertical direction y of the horizontal lines from their positions in the input image. The distortion graph would then be a plot of horizontal offset dx as a function of vertical distance y along the line, or a plot of vertical offset dy as a function of horizontal distance x along the line . In each case the graph would have a maximum positive value of the offset and a maximum negative value of the offset, each of which can be identified. The program may then be arranged to correct the data points for lens distortion which will be a known quantity for the optical system, and then to determine the magnitude of the gross distortion. The magnitude may be defined in a number of ways, but is typically defined as the mean magnitude of the maximum positive and negative offsets dx between the distorted line and its original position. This can be determined for each of the lines and then averaged to determine an average value of gross distortion for the NVD. This value may then be compared with a threshold value, and if it exceeds the threshold the program may determine that the test has been failed, otherwise it may determine that it has passed. The pass/fail result or the mean distortion value, or both can be indicated to the user on the display 1 18. It will be appreciated that other straight features, such as edges, may be used instead of, or as well as, the lines described above .

Referring to Figures 13 and 14, the program may also be arranged to perform a discontinuity test. This may use the same input image as the gross distortion test, and the first steps of acquiring a sequence of images, averaging and normalising, locating the grid, converting to points on a distortion graph and correcting for lens distortion may be the same as described above for the gross distortion test. Then, the data points for each line may be scanned for any discontinuities. As there will be some noise in the image, a simple offset between two adjacent points on a line is not sufficient to indicate a discontinuity. The program may therefore be arranged to identify two sets of points each forming a substantially continuous vertical line 124a, 124b, i.e. having equal x coordinates, with an offset dx between the x coordinate of one of the two lines and the x coordinate of the other, and with the two lines being offset from each other in the y direction so that shifting one of them in the x direction by the offset dx would form a single continuous line . Once these two lines 124a, 124b have been located, the value of the offset dx between them can be determined and this represents the magnitude of the discontinuity. This process may be carried out for all of the lines on the grid 122 and a total number of discontinuities with their magnitudes calculated. These magnitudes may be displayed to the user on the screen 1 18, or they may be analysed by the program against a set of criteria and a simple pass or fail result determined which may then be displayed. It will be appreciated that other straight features, such as edges, may be used instead of, or as well as, the lines described above.

Referring to Figure 15, the program may further be arranged to perform a magnification test on the NVD . Ideally the output image scale should be equal to that of the input image, i.e. a magnification of 1 : 1 , and this should be the same for both sides of the NVD if it is a binocular device. In order to perform this test the program may be arranged to acquire a sequence of images using an input image having large features, such as the circles in the plain white input image or the large bars in the resolution test image. The sequence of images may then be averaged and normalised as in other tests described above. The program may then be arranged to locate one or more large features in the output image, such as the circles or large bars, and to compare one or more dimensions of those features with corresponding dimensions from a reference image or from a reference values of the dimensions. It may then be arranged to calculate the magnification of the NVD, for example as a ratio or percentage, and compare the value of the magnification with a threshold value to determine whether the system passes or fails the test. The results may then be displayed on the screen 1 18, either as a pass/fail indication or as a value of the magnification, or both. This may be repeated for both sides of the NVD if appropriate .

Referring to Figure 16 the program may further be arranged to perform an alignment test to check that the two sides of a binocular NVD are aligned with each other. This test may use the same input image as is used for a manual test of alignment, which is in fact two images, one for the left hand side and one for the right hand side of the NVD . The test kit 100 may be arranged to generate these two images separately, one after the other. One of the images may have a square box and the other may have a cross, with the cross and the square box being in corresponding places in the two images, so that, if the two sides of the NVD are aligned, the cross and the box will be in corresponding places in the output images from the two sides. The apparatus may include a bridge piece which is arranged to be placed over two eyepieces of the NVD and to combine the images from them into a single image. If the test kit 100 only generates a single image, the NVD may be mounted on a slider so that the NVD can be moved between a first position in which the LHS of the NVD is located over the input image, and a second position in which the RHS of the NVD is located over the input image. The program may be arranged to acquire a first sequence of images from the LHS of the NVD using the box as the input image, and then to average the images and normalise the averaged image, and to calculate the location of the box in the image . It may then then arranged to acquire a second sequence of images from the RHS of the NVD using the cross as the input image, and again to average and normalise and calculate the location of the cross in the image. It is then arranged to calculate whether the cross and box in the respective output images are aligned such that, if the images were overlaid, the cross would be within the box in both the horizontal (x) and vertical (y) directions. If it would, then the program may be arranged to display an indication that the test has been passed, but if not then the program may be arranged to display an indication that the test has been failed. Alternatively, the test kit 100 may be arranged to generate two output images simultaneously on spaced-apart screen areas so that each of the two sides of a binocular NVD can be used to image a respective one of the images. The bridge piece may then output a single combined image of, for example, the square and the cross, with the image from the two sides superimposed on each other. The program may then analyse the single combined image to determine the degree of misalignment between the two parts of the combined image, for example in two orthogonal directions, and determine from those measured misalignments whether the alignment test is passed or failed.

Referring to Figure 17 the program may further be arranged to perform an image stability test. This is arranged to test for flicker in the image and comprises taking a number of images in a sequence with a high frame rate, for example in the range 100 to 200Hz, calculating an average grey level for all or part of each image in the sequence, and measuring the variation in the average grey level between the images. The area of the image used may be a small part of the image, so as to keep the data acquisition rate at a manageable level, and may be in a central region of the image which is generally brighter. The variation may be measured by calculating the variance of the average grey levels of the sequence of images. The variance may be compared with a threshold value, and if it is above the threshold the stability test may be determined as failed, and if it is below, the test may be determined as passed. The pass/fail result may be indicated to the user on the screen 1 18. Referring to Figure 18 the program may further be arranged to perform a signal to noise ratio test on the image. For this test, the input image that the test kit 100 is arranged to generate is a 0.2mm diameter spot with a luminance of lxl 0^"5 foot candles having the spectral content of 2856±50K blackbody radiation. The program is arranged to capture a sequence of output images from the NVD using that input image with a frequency of 10 frames per second. The program is then arranged, for each image, to measure the measure the average signal S₀ from the camera over the 0.2mm diameter area of the bright spot, the standard deviation N₀ of the signal S₀ over the sequence of images, the average signal S_bkd from the camera over a 0.2mm diameter area outside the bright spot, which is taken to be the background signal, and the standard deviation N_bkd of the background signal S_bkd- The program is then arranged to calculate the signal to noise ratio (SNR) of the NVD using the calculation:

SNR = (So - Sbkd)/K(N₀ ² - Nbkd²)' It will be understood that this corresponds to the standard MIL-SPEC SNR calculation as defined in MIL-PRF-49428(CR).

Once the SNR has been calculated, this is compared with a threshold value, and a pass/fail indication provided to the user.

While the systems shown in Figure 1 is only arranged to test a single output image at one time, as mentioned above, in a modification to this the test set 100 is arranged to generate two images simultaneously, and there are two cameras 106, one for mounting on each side of a binocular NVD, and both of which are connected to the tablet 1 12 at the same time . This allows the simultaneous testing of both sides of the binocular device, and also means that, for the test that require images from both sides, such as the alignment test and the magnification and gain tests, the input images for the two sides can be coordinated and the output images detected and processes simultaneously.

Referring to Figure 19, in a further modification, rather than a separate tablet, the processing may be carried out entirely within the test set 100a. In this case the test set 100a may be arranged to generate all of the required input images in sequence, with one image for each side of the binocular device 104a generated simultaneously, and two cameras 106a, 106b one for mounting on each side of the NVD to capture the required output images for all, or any combination, of the test described above to be performed in an automated cycle. The test set 100a may include a touchscreen screen 1 18a which may provide and or all of the functions of the tablet 1 12 described above. The output to the user may then be in of an overall pass/fail result, or a separate result for each test, or more detailed data from each test as described above.

Referring to Figure 20, in a further aspect of the modification of Figure 19 the test set 100a may include an active LCD screen 150a, which may be arranged over a light source 152a, and which may be controlled by the program so as to generate different images as required. These may include the plain white image (with or without circular regions marked), the square grid pattern, or the resolution test bar image, or any other images that may be used for testing the NVD 104a. The light source 152a may comprise a lamp 154a located inside an integrating sphere 156a and a diffuser 158a. One or more lenses 160a may be provided over the LCD screen 150a to provide some focussing of the images generated by the LCD screen. This image generation system allows high flexibility and automation of the various tests described above. It will be appreciated that other methods of generating the image such as an LED or OLED display or a micro-mirror array can be used.

Claims

1. A system for testing a night vision device, the system comprising detection means arranged to detect an image produced by the night vision device and to output image data encoding the image, and processing means arranged to analyse the image data to determine whether the image meets at least one criterion, and to generate an output indicative of whether the at least one criterion is met.

2. A system according to claim 1 wherein the processing means is arranged to identify the presence of at least one feature in the image.

3. A system according to claim 1 or claim 2 further comprising a display screen.

4. A system according to claim 3 arranged to provide an indication on the display screen of whether the at least one feature is detected.

5. A system according to claim 3 or claim 4 arranged to display the image on the display screen.

6. A system according to claim 5 arranged to highlight the feature on the image.

7. A system according to any preceding claim further comprising image generating means arranged to generate an input image so that the image produced by the night vision device is an output image reproducing the input image .

8. A system according to claim 7 wherein the input image is a plain image and the processing means is arranged to identify features in the enhanced image which are not present in the input image.

9. A system according to claim 8 wherein the processing means is arranged to identify features meeting at least one criterion, and to determine the number of such features in at least an area of the image .

10. A system according to claim 8 or claim 9 wherein the processing means is arranged to control the display so as to highlight each of the features that meet the at least one criterion.

1 1. A system according to any one of claims 7 to 10 wherein the input image comprises a plurality of features of the same shape but of different sizes or orientations, and the processing means is arranged to determine which of the corresponding features in the output image meet the at least one criterion.

12. A system according to claim 2 or any claim dependent thereon wherein the feature is a blemish or the features are blemishes.

13. A system according to claim 2 or any claim dependent thereon where the feature is a hexagonal pattern, or the features are parts of a hexagonal pattern.

14. A system according to any one of claims 7 to 1 1 wherein the processing means is arranged to measure a parameter of the output image and to calculate from the parameter the gain of the device.

15. A system according to claim 14 wherein the parameter is a measure of brightness of at least a part of the output image .

16. A system according to any preceding claim wherein the processing means is arranged to measure a parameter of the image or the output image, and to calculate from the parameter the level of gross distortion produced by the device.

17. A system according to claim 16 wherein the parameter is a measure of offset between a curved feature in the image or the output image and a reference straight feature.

18. A system according to any preceding claim wherein the processing means is arranged to detect discontinuities produced by the device in the image or the output image.

19. A system according to claim 18 wherein the processing means is arranged to identify in the image or the output image two parallel straight features and to measure the offset between the two straight features.

20. A system according to any preceding claim wherein the processing means is arranged to measure the magnification of the device .

21. A system according to claim 20 wherein the processing means is arranged to measure a dimension of a feature in the image or the output image and to compare the dimension with a reference value of the dimension to determine the magnification of the device.

22. A system according to any preceding claim wherein the processing means is arranged to measure alignment between two sides of a binocular night vision device .

23. A system according to locate a feature in an output image from each side of the binocular device and to compare the locations so as to determine the alignment of the two sides.

24. A system according to any preceding claim wherein the processing means is arranged to measure flicker in the image or the output image .

25. A system according to claim 24 wherein the processing means is arranged to measure a variation in brightness of at least a part of the image or the output image between frames acquired at a frame rate of at least 50Hz

26. A system according to any preceding claim wherein the processing means is arranged to measure signal to noise ratio in the image or the output image .

27. A system for aiding a user in focusing an optical device, the system being arranged to acquire a sequence of images output by the device, calculate a focus metric for each of the images, and generate a user-perceptible output indicative of the focus metric, wherein the output is arranged to vary in real time during adjustment of the focus of the device .

28. A system according to claim 27 wherein the output is a plot of the metric against time.

29. A system according to claim 27 or claim 28 which is arranged to identify a maximum value of the metric as the device passes through a point of maximum focus, and then to detect when the value of the metric returns to that maximum value and in response to generate a further user perceptible output.