WO2016037236A1 - Particle streak velocimetry method and apparatus - Google Patents

Abstract

Apparatus for performing particle streak velocimetry, the apparatus including: a radiation source that generates electromagnetic radiation for exposing particles in a fluid flow; a sensor that senses radiation from the particles; and a control system coupled to the sensor and the radiation source, wherein the sensor control system: causes the radiation source to generate radiation having an intensity that varies continuously over an exposure time period; receives signals from the sensor, the signals being indicative of radiation from the particles during the exposure time period; and generates an indicator indicative of at least a direction of travel of particles in the fluid flow.

Description

PARTICLE STREAK VELOCIMETRY METHOD AND APPARATUS Background of the Invention

[0001] This invention relates to a method and apparatus for performing Particle Streak Velocimetry (PSV).

Description of the Prior Art

[0002] The reference in this specification to any prior publication (or information derived from it), or to any matter which is known, is not, and should not be taken as an acknowledgment or admission or any form of suggestion that the prior publication (or information derived from it) or known matter forms part of the common general knowledge in the field of endeavour to which this specification relates.

[0003] Particle Image Velocimetry (PIV) is a method of visualising fluid flows by measuring the movement of multiple particles within the flow and is used extensively in the fields of mechanical and aeronautical engineering. PIV allows the velocity of particles in a fluid flow to be captured, which in turn allows features of the flow itself to be quantified. In this regard, the governing equations of fluid mechanics are chaotic in nature and cannot be reliably simulated or simplified. As a consequence, in order to effectively design technology operating in fluids (such as aircraft, automotive vehicles, seacraft, building structures, human-powered vehicles, wind turbines, missiles/rockets/ordnance, etc.), or utilising internal fluids (such as in cooling systems, fuel delivery systems, exhaust vents, sprays, jets, etc.) it is necessary to measure the behaviour of flows experimentally.

[0004] The method of PIV has undergone extensive development over the past 25 years, and is well established as a highly reliable industry standard. PIV systems work by 'seeding' a fluid flow with luminescent or light reflecting particles, which are then illuminated by a laser sheet, providing a visual representation of a 2-dimensional plane in the flow. Traditionally, this has been achieved using lasers that generate a brief pulse of illumination, of the order of a few nano-seconds duration, with two lasers being used in synchronisation to produce respective pulses separated by a short period of time. Special high-speed PIV cameras are set up perpendicular to the illuminated plane and, controlled by a hardware system, take photographs in quick succession, with the resulting images showing the position of the seeding particles at respective times. By comparing the images, the distance travelled by each can be established, with this being used to derive the velocity of the particles. However, equipment of this form is typically expensive and difficult to operate accurately, the image processing algorithm requires considerable computation resources and can result in significant errors, thereby limiting the use of the process.

[0005] Several alternative methods for visualising fluid flows by measuring the movement of multiple particles within the flow have been proposed.

[0006] JP-S-5757265 describes a mechanism for measuring the speed and the direction of flow by illuminating particles suspended in fluid and modifying the intensity of the illuminating light so that the light intensity at the end of exposure is brighter than at the start of the exposure, thus the flow trace of the terminating point is photographed more brightly than that at the starting point. A similar process is described in Wung, T. S., Tseng, F. G., 1992; "A Color-coded Particle Tracking Velocimeter with Application to Natural Convection", Experiments in Fluids 13, pp. 217-223, in which the same result is achieved, albeit through the sensitivity jump in the detector.

[0007] However, these techniques only discriminate the starting and end points of the traces and do not allow the recovery of partly overlapped vectors in the cases of overseeded, or highly transient/fluctuating flows, since there are only two light intensity levels per image, namely a lower level at the beginning and a higher level at the end of the process.

[0008] Khalighi, B., Yong H. Lee, 1989, "Particle Tracking Velocimetry: An Automatic Image Processing Algorithm", Applied Optics, Vol 28, No. 20, pp. 4328-4332 proposed a method for coding the streaks asymmetrically with a light chopper so that the particle trajectory includes series of dots and streaks. Afterwards, the images are analysed to yield graphical direction information in a vector form. The method is accurate for images with comparatively straight streaks. However, when streaks with severe curvature occur, the accuracy decreases. This method also does not allow the recovery of the partly overlapped vectors.

[0009] DE19737933A and Gbamele et al.: "A method for validating two-dimensional flow configurations in particle streak velocimetry" TRANSACTIONS OF THE ASME. JOURNAL OF FLUIDS ENGINEERING, JUNE 2000, AS ME, U.S., vol 122, No. 2, pages 438-440, XP000997966 each describe a method for polychromatic PSV. The method uses three monochromatic adjoining laser light sheets of different wavelengths with a homogeneous power density distribution. The source of light is an Ar-Kr laser emitting in all three lines with a power of about 4 W. Similarly US-6, 549,274 describes the use of two- dimensional images that are recorded in a frequency-selective or frequency-band-selective manner, from which the flow is determined. The illuminating device for this purpose generates at least two, at least approximately parallel light sheets, arranged in spatial succession, generated in temporal succession, having electromagnetic waves of different frequencies or different frequency spectrums, which scan the detection space at least in areas. However, these methods are technically complicated and do not allow the mapping the flow in a vector form.

[0010] US-5, 812,248 describes a process and apparatus for generating object traces representing the movement of objects through a measuring space. The measuring space can be illuminated by a laser or other light source, or alternatively the objects can be self- illuminating. In either event, light from the objects is detected over a finite acquisition time, to generate the object traces. At an end of the acquisition time, one of several parameters that govern the object trace width is altered to increase the width, thereby giving the object trace the form of an arrow or vector. The altered parameter can be the brightness of illumination, the sensitivity of the image detector, or the position or state of system components, in particular the image detector, a lens or other active optical component between the objects and the detector, or an optical component between the light source and the objects. However, this process does not yield data that can be later digitised.

[0011] US2002/0145726 describes a Low-Cost alternating colour-pulse digital particle- image velocimetry DPIV (Digital Particle Image Velocimetry) system. The system uses a continuous-wave laser in mixed mode, a CCD (Charge Couple Device) camera, a PMT (Photo-Multiplier Tube), an image-processing card, a PC (personal computer) etc., with an add-on alternating-colour planar laser-sheet generating facility to achieve the purpose of DPrV (Digital Particle Image Velocimetry) of planar velocity measurements. With the addon facility, the laser beam from the continuous-wave laser operated in mixed mode is turned into a planar laser sheet with alternating colour at a designated frequency. The CCD (Charge Couple Device) camera captures the alternating-colour images of the flow field seeded with small particles. The images are then sent to a personal computer for analysis of the magnitude and direction of the velocity distribution of the flow field. In this method, the first illumination pulse occurs in a different colour than the second, which upon an analysis allows the direction of the trace to be calculated. This relies on mechanical moving parts in the form of a system of rotating filters, making such methods technically complicated and can significantly worsen the timing accuracy.

[0012] US2013/0242286 describes an imaging device can take an image of a flow field including tracer particles includes a compound-eye lens formed from a large number of monocular lenses, which take images of images taken by an imaging lens. Each of the multiple monocular lenses functions as one imaging device. This can enable measurement precision to be enhanced by suppressing the influence of ghost particles, while reducing the equipment cost by minimizing the number of imaging devices. A space for installing the imaging device can be easily ensured. If a large number of imaging devices are used, not only do they require time and manpower for setting up, but there is also a possibility that the measurement precision will be degraded due to displacement of an axis of the imaging devices caused by vibration, etc. When the imaging device having the compound-eye lens is used, setup is simplified, and measurement precision can be ensured. However, again this is complex and expensive.

[0013] Accordingly, existing methods for visualising fluid flows by measuring the movement of multiple particles within the flow are either too technically complicated and/or have a limited accuracy and ability for further image processing.

Summary of the Present Invention

[0014] In one broad form the present invention seeks to provide apparatus for performing particle streak velocimetry, the apparatus including:

a) a radiation source that generates electromagnetic radiation for exposing particles in a fluid flow;

b) a sensor that senses radiation from the particles; and,

c) a control system coupled to the sensor and the radiation source, wherein the control system: i) causes the radiation source to generate radiation having an intensity that varies continuously over an exposure time period;

ii) receives signals from the sensor, the signals being indicative of radiation from the particles during the exposure time period; and,

iii) generates an indicator indicative of at least a direction of travel of particles in the fluid flow.

[0015] Typically the intensity varies at least one of:

a) progressively over the exposure time period; and,

b) increases throughout the entire exposure time period.

[0016] Typically the control system includes at least one electronic processing device that processes the signals from the sensor.

[0017] Typically the electronic processing device generates an image showing streaks corresponding to the moving particles, and wherein the intensity of the streak varies in accordance with the varying intensity to thereby indicate a direction of movement.

[0018] Typically the electronic processing device generates a vector map indicative of the flow of particles in the fluid, the vector map including at least one vector indicative of a magnitude and direction of movement of at least one particle within the fluid flow.

[0019] Typically the electronic processing device derives the vector map from an image.

[0020] Typically the electronic processing device generates the vector map by:

a) identifying intensity peaks within the signals received from the sensor;

b) identifying ridges using the intensity peaks; and,

c) analysing at least one ridge to determine a vector indicative of at least one of a direction and magnitude of particle movement.

[0021] Typically the electronic processing device:

a) generates an image intensity map;

b) removes background noise from the intensity map; and,

c) identifies intensity peaks in the intensity map. [0022] Typically the electronic processing device removes the background noise using at least one of:

a) an intensity threshold; and,

b) a frequency-domain filter.

[0023] Typically the electronic processing device:

a) enhances the intensity peaks using a filter;

b) performs edge detection to identify ridge edges; and,

c) generates a ridge map using the ridge edges.

[0024] Typically the electronic processing device performs edge detection using at least one of:

a) a neighbour clustering algorithm; and,

b) an intensity gradient analysis.

[0025] Typically the electronic processing device:

a) performs ridge isolation to identify individual ridges; and,

b) generates a vector for each ridge using the ridge intensity.

[0026] Typically the electronic processing device:

a) identifies a maximum intensity for each ridge, the maximum intensity corresponding to a vector end; and,

b) identifies a ridge point furthest from the maximum intensity as a vector start.

[0027] Typically at least one electronic processing device:

a) determines changes in intensity along a vector;

b) compares the changes in intensity to the varying intensity of radiation generated by the radiation source; and,

c) uses a result of the comparison to determine at least one of:

i) components of movement at least one of:

(1) perpendicular to a sheet of radiation used to expose the particles; and,

(2) parallel to the sheet; and,

ii) overlapping streaks;

iii) partly visible streaks; and, iv) curved streaks.

[0028] Typically the radiation source includes a continuous intensity radiation source and a modulator for varying the intensity of the radiation.

[0029] Typically the radiation source is a continuous wave laser and the modulator is an acousto-optic modulator.

[0030] Typically the control system includes:

a) a pulse generator that generates an electrical control signal including at least one pulse having a defined pulse profile; and,

b) a driver coupled to the pulse generator that drives the acousto-optic modulator in accordance with the pulse profile of the control signal.

[0031] Typically the control system includes a trigger unit that generates a trigger signal to initiate the exposure time period.

[0032] Typically the trigger signal synchronises generation of the electromagnetic radiation of varying intensity and capturing of signals from the sensor.

[0033] Typically the sensor is a charge-coupled device sensor.

[0034] Typically the radiation source generates a substantially planar sheet of radiation, and wherein the image sensor is positioned substantially offset to and facing the sheet.

[0035] In another broad form the present invention seeks to provide a method for performing particle streak velocimetry, the method including:

a) exposing particles in a fluid flow to electromagnetic radiation from a radiation source, the radiation having a continuously varying intensity over an exposure time period;

b) using a sensor to sense radiation from the particles; and,

c) using signals received from the sensor during the exposure period to generate an indicator indicative of at least a direction of travel of particles in the fluid flow. [0036] In another broad form the present invention seeks to provide apparatus for generating a vector map indicative of the flow of particles in a fluid, the apparatus including an electronic processing device that:

a) identifies intensity peaks within an image derived from radiation from particles in a fluid flow, the particles being exposed to electromagnetic radiation having a continuously varying intensity over an exposure time period;

b) identifies ridges using the intensity peaks; and,

c) analyses at least one ridge to determine a vector indicative of at least one of a direction and magnitude of movement of a respective particle.

[0037] Typically the electronic processing device:

a) generates an image intensity map;

b) removes background noise from the intensity map; and,

c) identifies intensity peaks in the intensity map.

[0038] Typically the electronic processing device removes the background noise using at least one of:

a) an intensity threshold; and,

b) a frequency-domain filter.

[0039] Typically the electronic processing device:

a) enhances the intensity peaks using a filter;

b) performs edge detection to identify ridge edges; and,

c) generates a ridge map using the ridge edges.

[0040] Typically the electronic processing device performs edge detection using at least one of:

a) a neighbour clustering algorithm; and,

b) an intensity gradient analysis.

[0041] Typically the electronic processing device:

a) performs ridge isolation to identify individual ridges; and,

b) generates a vector for each ridge using the ridge intensity.

[0042] Typically the electronic processing device: a) identifies a maximum intensity for each ridge, the maximum intensity corresponding to a vector end; and,

b) identifies a ridge point furthest from the maximum intensity as a vector start.

[0043] Typically at least one electronic processing device:

a) determines changes in intensity along a vector;

b) compares the changes in intensity to the varying intensity of radiation generated by the radiation source; and,

c) uses result of the comparison to determine components of movement at least one of:

i) perpendicular to a sheet of radiation used to expose the particles; and, ii) parallel to the sheet.

[0044] In another broad form the present invention seeks to provide a method for generating a vector map indicative of the flow of particles in a fluid, the method including:

a) identifying intensity peaks within an image derived from radiation from particles in a fluid flow, the particles being exposed to electromagnetic radiation having a continuously varying intensity over an exposure time period;

b) identifying ridges using the intensity peaks; and,

c) analysing at least one ridge to determine a vector indicative of at least one of a direction and magnitude of movement of a respective particle.

Brief Description of the Drawings

[0045] An example of the present invention will now be described with reference to the accompanying drawings, in which: -

[0046] Figure 1 is a schematic diagram of an example of apparatus for performing Particle Streak Velocimetry;

[0047] Figure 2 is a flow chart of an example of a method for performing Particle Streak Velocimetry;

[0048] Figures 3A and 3B are examples of positive and negative images of streaks left from white dots printed on a black rotating disc being exposed to gradually increasing radiation;

[0049] Figures 3C and 3D are examples of enlarged portions of the images of Figures 3A and 3B respectively; [0050] Figures 3E and 3F are examples of positive and negative images of streaks left from particles seeded in a fluid flow around a cylinder, being exposed to gradually increasing radiation and used to provide instantaneous vector map of the flow;

[0051] Figures 4A and 4B are examples of vector maps overlaid in the images of Figures 3C and 3D;

[0052] Figure 5 is a flow chart of an example of a method of generating a vector map;

[0053] Figure 6 is a schematic diagram of a second example of apparatus for performing Particle Streak Velocimetry;

[0054] Figure 7 is a schematic diagram representing the timing of operation of components of the apparatus of Figure 6;

[0055] Figure 8 is a flow chart of a second example of a method of generating a vector map;

[0056] Figure 9 is a schematic diagram of an example of an intensity map for the image of Figures 3C and 3D;

[0057] Figure 10 is a graph of an example of an intensity threshold;

[0058] Figure 11 is a schematic diagram of an example of the intensity map of Figure 9 after thresholding;

[0059] Figure 12 is a schematic diagram of an example of an enhanced version of the intensity map of Figure 11 ;

[0060] Figure 13 is a schematic diagram of an example of a neighbourhood analysis grid used in analysing the intensity map;

[0061] Figure 14 is a schematic diagram of an example of detected edges in the enhanced intensity map of Figure 12;

[0062] Figure 15 is a schematic diagram of a colourised ridge map;

[0063] Figure 16 is a schematic diagram of an isolated ridge;

[0064] Figure 17 is a schematic diagram illustrating ridge intensity maxima detection;

[0065] Figure 18 is a graph of ridge edge distance from the intensity maxima;

[0066] Figure 19 is a schematic diagram of vectors overlaid on the ridge map; and,

[0067] Figure 20 is a schematic diagram of a vector map.

Detailed Description of the Preferred Embodiments

[0068] An example of an apparatus and a method for performing Particle Streak Velocimetry (PSV) will now be described with reference to Figures 1 and 2. [0069] In this example the apparatus 100 includes a radiation source 110, a sensor 120 and a control system 130 coupled to the sensor 120 and the radiation source 110.

[0070] In use, the radiation source 110 generates electromagnetic radiation for exposing particles in a fluid flow F. The electromagnetic radiation is typically visible radiation generated by a laser, although this is not essential and other wavelengths of radiation and other sources could be used depending on the preferred implementation. The sensor 120 senses radiation from the particles and in this regard, the particles are typically at least partially reflective to or luminescent in response to the radiation generated by the radiation source 110.

[0071] At step 200, the control system 130 causes the radiation source 110 to generate radiation having an intensity that varies continuously over an exposure time period, for example by triggering a modulator that modulates radiation output from a constant intensity radiation source.

[0072] Simultaneously, at step 210, radiation reflected or emitted from the particles is sensed during at least the exposure time period. This can be achieved by triggering the sensor 120 and/or a processing device that captures signals from the sensor, as will be described in more detail below. The sensed radiation can then be used to generate an indicator indicative of at least a direction of travel of particles in the fluid flow.

[0073] The indicator could be of any appropriate form and in one example is an image including streaks corresponding to the moving particles, with the intensity of the streak varying in accordance with the varying intensity of the exposing radiation, to thereby indicate a direction of movement. An example image taken of white dots printed on a rotating black disc is shown in Figures 3A and 3B, with close up views shown in Figure 3C and 3D. As shown in these examples, the images include streaks, each of which corresponds to a single dot printed on the disk.

[0074] In this example, the exposing radiation increases in intensity throughout the exposure time period, meaning that the streaks demonstrate a corresponding increase in intensity towards their end. Accordingly, the streaks include information regarding the direction of dot movement, with the direction of movement being from the low intensity to the high intensity end of the streak, as well as speed information based on the length of the streak and the duration of the exposure time period. Thus, this allows the velocity of particles to be easily determined solely through visual inspection of the image.

[0075] By way of example, in the images of Figures 3A and 3B it is possible to determine the direction of rotation of the disc, namely clockwise, with the linear velocity of the particles increasing towards the outer edge of the disc, as would be expected.

[0076] A further example is shown in Figures 3E and 3F. In this example, fluid having particles entrained therein is flowing past a cylinder C, with a number of features of the resulting flow being clearly visible, such as areas of stagnation S, eddies E and flow F. As will be understood by persons skilled in the art.

[0077] Accordingly, by generating an image, this provides a technique for allowing gross features of the flow, as well as the speed and direction of individual particles, to be easily visualised.

[0078] Additionally and/or alternatively the indicator can be in the form of a vector map, which includes at least one vector indicative of a magnitude and direction of movement of at least one particle within the fluid flow. An example vector map is shown in Figures 4A and 4B, with a method for generating the vector map being described in more detail with reference to Figure 5.

[0079] For the purpose of this example, it is assumed that the generation of the vector map is performed at least in part using an electronic processing device forming part of a processing system, such as a suitably programmed computer system or the like, as will be explained in more detail below.

[0080] In this example, at step 500 an image is acquired. The image is of radiation reflected or emitted from particles in a fluid flow, with the particles being exposed to electromagnetic radiation having a continuously varying intensity over an exposure time period. The image could be acquired using the apparatus and method described above with respect to Figures 1 and 2, or other similar techniques. Alternatively, a previously created image could be retrieved from memory, a database, or the like, or received from a remote computer system, server, or the like. [0081] At step 510, the electronic processing device identifies intensity peaks within the image. The intensity peaks can be identified using any suitable image processing technique, such as by scaling the image intensity and then identifying portions of the image having an intensity greater than a threshold, as will be described in more detail below.

[0082] At step 520, the electronic processing device identifies ridges using the intensity peaks, for example by performing an edge detection technique or the like. Following this, at step 530, the electronic processing device analyses at least one ridge to determine a vector indicative of at least one of a direction and magnitude of movement of a respective particle.

[0083] Accordingly, it will be appreciated that this provides a straightforward technique for automatically processing images to generate vector maps, which can then be used in analysing fluid flow or other particle motion in more detail.

[0084] In any event, it will be appreciated that in the above examples, by varying the intensity continuously over the exposure time period, this has the benefit of allowing the direction of travel to be more easily identified. In particular, this allows a vector map to be more easily generated and further allows additional information regarding the direction of movement to be determined as will be described in more detail below. This can also be achieved using straightforward apparatus, allowing this to be deployed in a wide range of situations, as will be described in more detail below.

[0085] A number of further features will now be described.

[0086] The intensity of the radiation exposing the particles typically varies progressively over the exposure time period and more typically increases at a steady rate throughout the entire exposure time period. This makes determination of the direction of movement straightforward for both visual inspection of images, as well as computer based analysis for the determination of vector maps.

[0087] The control system typically includes at least one electronic processing device that processes the signals from the sensor. The electronic processing device can form part of a processing system, such as a suitably programmed computer system or the like. In one example, the computer system is a standard processing system such as a 32-bit or 64-bit Intel Architecture based processing system, which executes software applications stored on non- volatile (e.g. , hard disk) storage, allowing for the image analysis to be performed. However, this is not essential, and alternatively the electronic processing device could be any suitable electronic processing device such as a microprocessor, microchip processor, logic gate configuration, firmware optionally associated with implementing logic such as an FPGA (Field Programmable Gate Array), or any other electronic device, system or arrangement.

[0088] When generating the vector map, the electronic processing device typically generates an image intensity map, removes background noise from the intensity map and identifies intensity peaks in the intensity map. In this regard, the electronic processing device can remove the background noise using an intensity threshold, for example by removing anything below the threshold, although other suitable techniques could be used. Alternatively, the electronic processing device may remove the background noise using frequency-domain methods. For example, frequency-domain filters may be used to reduce both regular and irregular noise. Further details of suitable techniques for removing background noise (and other background images) will be described in due course.

[0089] In order to facilitate the detection of the ridges by the electronic processing device, the radiation source 110 may be programmed to generate radiation, starting the pulse with a preset initial value that is above the zero level. Edge detection could be performed using any appropriate algorithm, but in one example this is achieved using neighbouring clustering algorithm, as will be described in more detail below. Alternatively, the edge detection could also be achieved by analysis of intensity gradient, examples of which will also be described further below.

[0090] The electronic processing device then performs ridge isolation to identify individual ridges and uses at least some of the ridges to generate a vector using the ridge intensity. As part of this, the electronic processing device typically identifies a maximum intensity for each ridge, the maximum intensity corresponding to a vector end and identifies a ridge point furthest from the maximum intensity as a vector start. Whilst the generation of a single vector can be useful for allowing a velocity analysis of a flow, in some examples a path of connected vectors may be calculated from each streak. This can allow for more accurate velocity vector calculation, and the ability to track curved streaks. For instance, the path may originate from the highest intensity point in the ridge, and an intensity gradient analysis may be used to create the path to the end of the streak. Further details of suitable techniques will be described in due course.

[0091] In addition to determining a vector representing the direction and magnitude of the velocity, the electronic processing device can further determine changes in intensity along the streak. These can then be compared to the varying intensity of radiation generated by the radiation source allowing this to be used to determine a number of features including for example a component of movement of the particles perpendicular to a sheet of radiation used to expose the flow, overlapping and partly visible streaks or curved streaks. In this regard, a vector will correctly present the direction and magnitude of the velocity of a particle in the flow only if the particle moves parallel to and within the plane of the light sheet used to expose the flow. Processing streaks that demonstrate deviations in the intensity of the emitted or reflected light along their lengths from the time varying intensity of the exposing radiation will result in generating erroneous vectors in respect to the motion of the particles within the interrogation plane. Processing curved or overlapped streaks will also result in generating erroneous vectors. The electronic processing device can be programmed to determine and to eliminate the erroneous vectors. However, variations in the intensity along the streak length (i.e. intensity gradient) and changes in the shape of the streak can be indicative of particles motion out of the interrogation plane and as such they can be quantified and used to provide information about the 3-dimensional motion of the flow. The use of intensity gradient along the streak typically requires that the path of the particle be resolved more accurately, for example with the general methods mentioned above for calculating a path of connected vectors.

[0092] A second example of apparatus for performing Particle Streak Velocimetry (PSV) will now be described in more detail with reference to Figure 6.

[0093] In this example, the apparatus 600 includes a radiation source having a constant intensity radiation source 610, such as a continuous wave laser and a modulator 611, such as an acousto-optic modulator, for varying the intensity of the radiation. An optic arrangement, such as a plano-concave lens 612 is also provided for converting the laser beam into a planar laser sheet 613 that exposes particles within the fluid flow, such as a water channel or a wind tunnel, with an image sensor 620, such as a charge-coupled device (CCD) sensor, being positioned substantially offset to and facing the sheet. The optic arrangement could also include fibre optic cables, allowing the illuminating radiation to be provided at a location remote from the radiation source 610 and modulator 611.

[0094] In this example, the control system includes an electronic processing device, typically forming part of a computer or other processing system 630, which receives signals from the sensor 620, and uses these signals to generate the image and/or vector map. The control system includes a pulse generator 631 that generates an electrical control signal including at least one pulse having a defined pulse profile and a driver 632 coupled to the pulse generator that drives the acousto-optic modulator 611 in accordance with the pulse profile of the control signal. The control system further includes a trigger unit 633 that generates a trigger signal to initiate the exposure time period, and in particular to trigger the pulse generator 631 and the processing system 630 and/or sensor 620 thereby synchronising generation of the electromagnetic radiation of varying intensity and capturing of signals from the sensor.

[0095] An example of the timing of the various components is shown in Figure 7. In this example, the trigger circuit 633 generates a square wave pulse, causing the pulse generator 631 to generate a triangular waveform, which in turn leads to the radiation 613 having an increasing intensity for a predetermined time period T. Meanwhile, the sensor 620 integrates sensed radiation over the time period Γ, as shown.

[0096] Accordingly, the above described apparatus utilises acousto-optic modulator (AOM) in which an oscillating electric signal drives a transducer, which is attached to a transparent material such as glass. The transducer's vibration creates sound waves in the glass that change the index of refraction, allowing an easy control of the intensity of the light exiting the AOM. The AOMs offer simple design and low power consumption (<3 watts), being much faster than typical mechanical devices such as tiltable mirrors. The time it takes an AOM to shift the transmitted beam is roughly limited to the transit time of the sound wave across the beam (typically 5 to 100 μ8). The intensity of the sound can be used to modulate the intensity of the light in the diffracted beam, which makes the AOM suitable for the purposes of the intensity coded process described above. [0097] In one example, the intensity of the radiation exiting the AOM can be varied between 15% and 99% of the input light intensity if the order of diffraction is zero, or it can vary between 0% and 80% if the order of diffraction is 1.

[0098] In use, the trigger unit 633 sends TTL (Transistor- Transistor Logic) compatible signals to the pulse generator 631 and the CCD Camera (via the processing system 630), with the pulse generator 631 producing a preset "triangular" electric signal which is transferred to the AOM analogue driver 632, which in turn drives the AOM. The AOM light output starts at a preset low intensity level and increases smoothly to the maximum, with the emitted radiation passing through the lenses 612 and illuminating the seeded flow. The CCD Camera 620 captures the light scattered from the tracers, with the images being stored by the processing system 630 and processed into a vector map.

[0099] The above described technique generates instantaneous graphical directional information about the flow. However, if required, the image acquisitions can be further processed by an electronic processing device to a vector form, which is more suitable for computational purposes.

[0100] The core algorithm for processing of captured images includes the following steps:

• Import of a single raw image or batch of raw images for processing by the electronic processing device;

• Analysis of each raw image to localise the summits (zones of maximal grey scale intensity, e.g. 255 in the 8-bit format) and the ridges. Vector positions coincide with the ridges, their magnitude equals the length of the ridges, and they are directed toward the summits (it should be noted that this is the case when using a single vector for each streak, but when using a series of vectors to follow the axis of a curved streak the sum of the vectors is directed to the "summits", rather than the individual vectors);

• Determination and elimination of the erroneous vectors and recovery of data from overlapped or partly visible streaks. As mentioned above, the criterion of validity of the vectors is based on the preset varying intensity of the radiation from the source of radiation, which provides data to recover bad vectors, by interpolation and extrapolation of the missed or overlapped sections of the streaks; • Collecting and processing information about the 3-dimensional motion of the flow by analysing variations in the intensity along the streaks length and changes in the shape of the streaks; and,

• Plotting the final velocity field in vector, contour or any other suitable format.

[0101] An example of the method of analysing images of a laser-illuminated fluid flow to calculate individual particle velocity and interpret the information as a vector field, thereby generating a vector map will now be described in more detail with reference to Figure 8.

[0102] For the purpose of illustration it is assumed that the computer system 630 implements the MATLAB programming environment, as matrices provide useful tools for digital image analysis and manipulation. However, it will be appreciated that this is not intended to be limiting and other suitable techniques can be used.

[0103] In this example, an image is initially acquired, with this preferably being in the form of an image file in an uncompressed or "lossless" compression format to avoid compression artefacts. The image is initially processed to convert this to an intensity map, IMG[], at step 800. This typically involves converting the image to a 2D matrix, as shown for example in Figure 9, with each pixel being represented by an intensity value ranging from 0 (black) to 255 (white).

[0104] At step 810 background noise cancellation is performed by applying an intensity threshold. In this regard, the matrix is swept, with the number of pixels of each intensity value being tallied in an array Idist[]. This is then plotted as shown in Figure 10, with the highest populations representing the greatest percent of the image, which in turn corresponds to the intensity of the background noise (which is largely of homogenous intensity). The highest value is then scaled down by a variable rangescale, to find an intensity corresponding to transition from background noise, to illuminated particle intensity, which is then stored in the variable I_base. IMG[] is then swept, all intensity values below I_base are set to zero - cancelling background noise, as shown in Figure 11.

[0105] At step 820, peak enhancement is performed to allow the peaks to be visualised. To achieve this a computationally efficient 'ridge' detection/error-checking method is used by applying a Gaussian low-pass filter to the matrix, as shown in Figure 12, which 'smooths' the data input (by use of weighted averaging) to exaggerate ridges (and converge background noise intensities) - aiding visualisation.

[0106] At step 830, edge detection is performed to differentiate ridges. In one example, this is achieved by grouping pixels into ridges, with the edgeFinder() function sweeping IMG[], and performing a 'neighbourhood' analysis. Fig 13 is a schematic of a neighbourhood analysis grid, labelling a centre pixel (C) surrounded by eight neighbouring pixels (N, S, E, W, NE, NW, SE, SW). As shown in Figure 13, each C pixel is analysed with the eight neighbouring pixels. If one of its neighbours is 0, it is flagged as an edge point. Also, if C pixel is on the image edge and non-zero, it is flagged as an edge (to deal with cut-off streaks). Each edge point is stored in the edgeList[] array, which is then distributed into the EDGES [] matrix, shown in Figure 14, and which is returned to imageConverter().

[0107] Each ridge corresponds to an individual particle (and therefore a resulting vector), and so must be treated individually. To isolate each ridge, the edgeScan() function sweeps the EDGES [] matrix until an edge is found, each point is checked off as being 'counted' in the ECOUNT[] matrix (to avoid rechecking pixels). When an edge is found (by a -1 value in EDGES []) the edgeRipple() function is used to find all the points of the ridge, the function works by applying a neighbourhood analysis to each pixel, flagging and recording all nonzero neighbours as members of the ridge (giving, and then applying the same process to those neighbours - the process is looped until all pixels in the ridge are located (when no unchecked neighbours exist). This method can be visualized as a ripple propagating across a body of water, spreading out as it travels - this is an efficient method for dealing with large matrices (as an entire sweep is not required). Each pixel is assigned a value in RIDGES [] which corresponds to the ridge number.

[0108] Isolating the pixels of each ridge by the above described edgeRipple() function utilises a method of region filling is often referred to as a 'boundary fill' algorithm, as it selects all pixels within a closed boundary. Boundary fill algorithms can only be used if the edges around each ridge are closed. If not, a flood fill algorithm could be included as an alternative, as discussed in further detail below.

[0109] At step 840 a ridge map is created by having individual RIDGES [] compiled to form RMAP[], a matrix with each ridge represented as a group of points of the same ridge number, as shown in Figure 15. Following this at step 850, ridge isolation is performed by loading IMG[], RMAP[] and EDGES [] are loaded into the vectorTransform() function. RMAP[] is swept, each ridge is moved to its own layer in the 3D matrix STACK[] - each layer is then analysed separately - layer by layer. Alternatively, the ridge may be moved to a sub-matrix (which may be more efficient from a memory management perspective compared to the use of a 3D matrix). The ridge is then 'windowed', extracted and placed into the VECTOR[] matrix, which is set to the correct size for each ridge so the ridges are isolated as shown in Figure 16, with the corresponding EDGES[] matrix being copied to the VEDGES[] matrix.

[0110] At step 860 vectorisation is performed. In this example, VECTOR[] and VEDGES[] are loaded into the FEA() function. The maximum intensity value is calculated, and scaled to 90% and set as I_max - this defines the peak intensity cut off. All peak intensities in VECTOR[] (above I_max) are used to calculate an average position, representing to the brightest point of the ridge, and corresponding to the end point of the particle's motion, as shown in Figure 17.

[0111] A polar sweep is executed from the end point to each edge point found in VEDGES[], the radius and theta values are stored in the thetaSweep[] array, as shown in Figure 18. The largest radius value corresponds to the point furthest away from the end point of the streak, which is the start point of the streak (assuming linear motion), the corresponding angle denotes the vector direction, the magnitude is a function of the radius. These are combined to construct a vector interpretation, as shown in Figures 19 and 20.

[0112] The above described method provides one example of suitable techniques that may be used to analyse images of the laser-illuminated fluid flow and subsequently generate the vector map, although it will be appreciated that a range of alternative techniques may be used to provide a similar outcome. Different image analysis techniques may be selected, depending on requirements, to increase the accuracy or versatility of the method. Although simplicity may be a major factor in the selection of techniques, more complex methods, such as machine-learning approaches to pattern-recognition, may also be considered.

[0113] A number of examples of possible alternative image analysis techniques will now be described, with regard once again to Figure 8 to provide context for the techniques within the method for generating the vector map. [0114] As discussed above, one approach for removing background noise from an acquired image of a laser-illuminated fluid flow involves converting the image to an intensity map as shown at step 800 and cancelling the background noise by applying an intensity threshold as shown at step 810. However, the background and noise removal may be effectively performed using alternative techniques, such as by using frequency-domain filters to reduce both regular and irregular noise.

[0115] It is noted that images with regularly distributed noise above the threshold intensity for background subtraction will generate erroneous results. To eliminate this noise, a discrete 2-D Fast Fourier Transform (FFT) method can used to isolate the regular noise components in the frequency domain. After eliminating these components in the frequency domain, an inverse FFT can be applied to reconstruct the image. This removes the regular noise patterns, but also degrades the image by removing information.

[0116] In one example, the FFT filter may be applied using the following steps:

• The image is loaded and converted to a matrix of image intensity, IMG[];

• The 2-D FFT algorithm is applied, creating the matrix EVIG_FFT[];

• The quadrants of EVIG_FFT[] are swapped diagonally to shift the DC component of the FFT to the centre of the matrix to aid visualisation;

• A user may manually select regions of IMG_FFT[], representing frequency bands, to be eliminated from the image; or optionally an algorithm may be applied to automate the process;

• The selected regions are removed from EVIG_FFT[], and the quadrants are again diagonally swapped, back to the original orientation; and,

• An inverse 2D FFT algorithm is applied to IMG_FFT[] to reconstruct the original image, excluding the eliminated frequency bands.

[0117] For irregular 'salt-and-pepper' noise (sparse and randomly-distributed intensity variations), the noise appears as intensity impulses, and therefore will appear in the highest frequency bands in the FFT. Elimination of these components and reconstruction of the image using an inverse FFT will reduce this irregular noise, but will also indiscriminately remove the high-frequency image data and thus will degrade the image. Alternatively, a median filter can be applied to reduce salt-and-pepper noise. The median filter would replace every pixel with the median intensity of a set of the nearby pixel intensities.

[0118] In one example, the median filter may be applied in the following steps:

• The image is loaded and converted to a matrix of image intensity, IMG[]. A duplicate of IMG is also created, IMG_MED[], which is used to construct the filtered image.

• A user may select the size of the median filter in pixels, representing the size of the sampled local neighbourhood surrounding each pixel.

• For each pixel in IMG[], located at coordinates (i, j), the surrounding neighbourhood is sampled and the median intensity selected. The corresponding pixel in IMG_MED[], also located at (i, j), is replaced by this median intensity.

[0119] It will be appreciated that any other suitable known frequency-domain filtering techniques may also be used for noise cancellation in the methods discussed above.

[0120] The example edge detection technique as described above with regard to step 830 uses a manual neighbourhood analysis, whereby a pixel is flagged as being an edge if any adjacent pixels are below the background intensity threshold. Subtraction of the background below an intensity threshold can produce well-defined ridges with closed edges. However, if the background cannot be eliminated by an intensity level, for example when the background intensity is close to the intensity of the streaks, this technique may be ineffective. Instead, filtering techniques based on the intensity gradients can be applied to identify edges.

[0121] Accordingly, examples of two alternative edge detection algorithms utilising intensity gradient analysis will now be described.

[0122] The first edge detection example involves the use of intensity gradient operators. Convolving specific 3x3 kernel matrices over an image enables the 2-D intensity gradients to be calculated, whereby discontinuities in the image are indicative of edges. Intensity gradient operators of the known Sobel, Prewitt or Roberts edge detection methods can be used in this way to rapidly perform edge detection.

[0123] In one implementation, an intensity gradient operator method of edge detection may be applied in the following steps: • The image is loaded and converted to a matrix of image intensity, IMG[]. A second matrix of equal size is created, EDGES [];

• A user may select the specific operator to be applied, for example, the Sobel operator;

• The intensity gradients in each direction are approximated separately, by orienting the operator kernel along a given direction, and then convolving the operator kernel on IMG. This creates a new intensity gradient matrix for each direction, GRADIENT_X[] and GRADIENT_Y[] , equivalent to taking a partial derivative of intensity along each direction;

• The total combined intensity gradient of IMG[] is calculated as the vector magnitude of the combined gradient fields, GRADIENT_X[] and GRADIENT_Y[], each representing a vector component of the gradient. The magnitude of the gradient field is stored in the matrix GRADIENT[]; and,

• The user may select a threshold gradient magnitude to be considered an edge, such that all values in GRADIENT [] higher than this level will be flagged as an edge in EDGES [].

[0124] The second edge detection example involves the use Canny edge detection. The Canny edge detection method is an effective and well-known staged process which combines a Gaussian filter with the edge detection operators mentioned above to identify edges. The edges are then refined to main edges by a series of operations. One disadvantage of the Canny method is a displacement of the edge towards the region of maximum intensity, as a result of the Gaussian smoothing function.

[0125] In one implementation, the Canny edge detection method may be applied in the following steps:

• The image is loaded and converted to a matrix of image intensity, IMG[]. A second matrix of equal size would be created, EDGES [];

• A Gaussian filter is applied to smooth the image;

• The intensity gradient is calculated using the intensity-gradient operators detailed above, producing the matrices GRADIENT_X[], and GRADIENT_Y[] . The vector magnitude of the intensity gradient is calculated and stored in matrix GRADIENT[], and the gradient direction is calculated using an arctangent function that preserves quadrant information, and stored in matrix GRADIENT_ANGLE[]. The angle is then rounded to vertical, horizontal, or 45-degree diagonals, and stored in the matrix GRADIENT_ANGLE_R[] ;

• A non-maximum suppression algorithm is applied to each pixel in GRADIENT[], to thin edges to a boundary line. For each pixel, if the value of GRADIENT[] is greater than both adjacent pixels along the direction specified by GRADIENT_ANGLE_R[], then the pixel is marked as an edge in EDGES [], otherwise it is suppressed; and,

• Hysteresis thresholding is applied to discard spurious edges. This may involve having a user select a minimum and maximum intensity gradient to be considered an edge. Any edge with a pixel with an intensity gradient above the maximum value is kept; otherwise it is suppressed in EDGES [].

[0126] Edges detected by these alternative methods may not produce closed boundaries. If a boundary-fill region filling algorithm is to be used, an additional step will be required to ensure ridges are completely contained. This could be applied by detecting when small breaks in non-closed boundaries are present, and flagging these as edges. Alternatively, a flood-fill algorithm could be used to identify ridge pixels, as discussed in further detail below.

[0127] Where ridges are defined by a closed edge boundary, a boundary-fill algorithm similar to the edgeRipple() function as described above will typically be used. However, the algorithm can be modified to adopt a 'scanline' algorithm, which fills intervals between points, rather than checking pixel-by-pixel.

[0128] In one example, a scanline algorithm may be applied in the following steps:

• The matrix containing flagged edge pixels, EDGES [], would be loaded. Streak edges partially located on the edge of the field of view would be suppressed. A matrix, RIDGES [], is created to flag pixels which are part of a streak, or 'ridge', numbering them uniquely for each ridge;

• Beginning at the top row of EDGES [], each row is scanned for intersections with an edge in EDGES []. Intersection points are numbered from left to right along each row;

• Each pair of intersection points, with an odd-numbered point on the left and an even- numbered point on the right, represents a line of pixels located inside a ridge, which are to be filled. Each pair of points is recorded in the stack, SEGMENTS []; • For each segment in SEGMENTS [], the corresponding pixels in RIDGES [] are set equal to a number unique to each ridge, and the adjacent row is checked for new segments which are added to the SEGMENTS [] stack; and,

• The process repeats until all streaks are filled.

[0129] Where closed edges cannot be created around each ridge, a 'flood fill' algorithm could be used. The flood fill algorithm would operate in a similar way to the boundary-fill algorithm discussed above, but would determine whether a pixel is part of a ridge by checking its intensity against a threshold, and adding it to the list of pixels if so. The threshold could be manually set, or could be based on analysis of the local image intensity, and the image intensity as a whole.

[0130] In the above discussed techniques for generating the vector map, the velocity vector calculation is based on an approximate method: the location of the vector end-point is calculated as a weighted average of the nearby pixels; and the vector start-point is defined as the pixel farthest away from the end-point. These vector calculation techniques may be improved in a number of ways as outlined below.

[0131] For example, vector calculation may utilise mapping of the pixels to a continuous field, enabling sub-pixel tracking resolution. Averaging the location of multiple pixels on a discrete grid requires sub-pixel resolution to correctly locate the centre-point. Mapping to a continuous coordinate system in the vectorisation step may enable the velocity vector to be calculated more accurately.

[0132] In another example, a variable intensity threshold may be provided for calculation of vector end-point location. The threshold defining the peak brightness may be user-variable, rather than fixed at 90%, as mentioned above in step 860.

[0133] Alternative start-point location calculation methods may be considered. The streak start-point should account for the light pattern scattered around the particle, caused by interference and diffraction - the farthest pixel may actually represent scattered light rather than the edge of the particle, which would artificially inflate the particle velocity. The start- point could be back-tracked from the end-point, enabling the particle path to be calculated in the same step, and enabling analysis of the intensity gradient. [0134] Alternative path tracking algorithms may also be considered. To enable curved streaks to be accurately vectorised, each streak would need to be converted to a connected path of vectors, rather than a single vector. This requires consideration of both the intensity gradient perpendicular to the particle motion, and the intensity gradient along the direction of particle motion, caused by the change in radiation intensity in time.

[0135] Two examples of methods to achieve vectorisation of curved particle paths will now be described.

[0136] The first vectorisation example involves a simple ridge tracking method. By calculating the first-order intensity gradient of each ridge, and beginning from the brightest point in the ridge, the particle path may be defined by following the shallowest intensity gradient, until reaching the edge of the ridge. This would produce a jagged representation of the path, which could then be smoothed by curve-fitting between the points, using a regression technique. The resulting curve would then be divided into linear segments approximating the curved path. Each segment would represent a vector on the map.

[0137] This method requires that the intensity gradient along the streak length is shallower than the intensity fall-off perpendicular to the particle path. In cases where the tracked particles are larger than one-pixel in diameter, the perpendicular gradient would be shallower, or zero, thus violating the requirements of the method. To track particles in this case an averaging method would be required, either of the image or resultant vectors, to reconstruct the average path along the direction of particle motion.

[0138] In one implementation, the simple ridge tracking method may be applied in the following steps:

• Each streak represented in RIDGES [] is isolated in a smaller matrix, RIDGE[];

• A user may manually input, or an algorithm would generate, the minimum intensity threshold for a pixel to be considered as part of the streak, as a function of the maximum intensity value in RIDGE[];

• Starting with the brightest pixel, and moving pixel-by-pixel along the direction of shallowest intensity gradient, a list of points is recorded until reaching the last connected pixel above the lower intensity threshold; • The recorded point and its adjacent pixels is analysed to create a weighted-average centroid position of the group, weighted by intensity. This offset centroid position is recorded for each point in the list, stored as a list of points in CENTROIDS[];

• Polynomial regression is applied to fit a line or curve to CENTROIDS[]. The intersection between the fitted line or curve and a bounding box containing all centroid points would define the start and end points of each streak; and,

• The velocity vector for each streak is given by the distance between the start and end points of the streak. Optionally, the fitted line or curve may be subdivided according to a user-selected, or automatically generated, number of segments, which each would represent a fraction of the exposure time. This would be required to be applied across the entire field, and would enable curved streaks, created from excessive exposure time, to be processed.

[0139] The second vectorisation example involves canal surface fitting. Vincenzo Caglioti, Alessandro Giusti 2009, "Recovering ball motion from a single motion-blurred image", Computer Vision and Image Understanding, Vol. 113, No. 5, pp. 590-597 (Caglioti et al. (2009)) showed that the path of a ball can be calculated in 3-D from a long-exposure image of its motion by fitting a 'canal surface' to the edges of the 'streak'. In that case the canal surface is the surface swept by the circular cross-section of the spherical ball along the path of motion: the image of the motion captures the side view of this canal surface.

[0140] Given the edges of the streak are well-resolved, the path travelled by the centre -point of the ball can be reconstructed by analysing 'coupled' points on the edges, by the method presented by Caglioti et al. (2009). This method may be applied by treating the tracked particles as spheres. However, the change in the apparent diameter of the sphere would be due to the intensity of illumination changing, rather than moving out-of-plane.

[0141] In one implementation, the canal surface fitting method may be applied in the following steps:

• Two curves, A and B, are fitted to the side edges of each streak, and the normal and tangent vector to the curves at each point are calculated; • Each point on curve A is compared against all points on curve B to detect whether any two points are coupled, meaning they exist on the same cross section of the canal surface, by the method presented in Caglioti et al. (2009);

• For each pair of coupled points, the midpoint located between the two points is calculated and stored in a list, MIDPOINTS [];

• Polynomial regression is applied to fit a line or curve to MIDPOINTS []. The intersection between the curve and a bounding box containing all midpoints would define the start and end points of each streak; and,

• The velocity vector for each streak would be given by the distance between the start and end points of the streak. Optionally, the fitted line or curve would be subdivided according to a user-selected, or automatically generated, number of segments, which each would represent a fraction of the exposure time. This would be required to be applied across the entire field, and would provide enable curved streaks, created from excessive exposure time or particle velocity, to be processed.

[0142] A limitation of this method is the requirement for accurate definition of the streak edges, as the tangent to the edge curves are used to reconstruct the trajectory. Also, the radius of curvature of the trajectory must be larger than the radius of the particles being tracked. Due to these limitations it may be preferable to use a simpler algorithm using gradient and intensity, but this method may nevertheless be advantageous in some circumstances.

[0143] In any event, the above described PSV techniques provide a method for performing visualisation of the flow, allowing a simple apparatus set up to be used, with a single, continuous wave, laser being used in conjunction with a AOM, whilst still allowing for presenting the resulting data in digital vectorised format. The PSV technique does not require the usage of specialised recording devices and virtually any digital camera can be used for recording the images. These factors significantly reduce the cost of the PSV apparatus compared with the cost of a PIV apparatus and at same time deliver enhanced capabilities.

[0144] The approach of Particle Streak Velocimetry (PSV) uses image streaks to determine velocity, generated by timed exposure of the illuminated seeded particles. The technique uses an acousto-optic modulator (AOM) to vary the intensity of the laser with time, allowing the start and end points of each streak to be established, and therefore allowing a vector field of velocity points to be constructed, performing the same task as PIV techniques.

[0145] The three major advantages of the new PSV technology over the existing PrV technology are that: (i) it only requires one laser; (ii) it only requires capturing a single image; and (iii) it has a potential for providing information regarding the 3-dimensional motion of the flow using the same simple set up. These advantages allow PSV systems to be built at a significantly lower cost compared with PrV systems.

[0146] A number of other advantages will now be described.

[0147] Current professional methods for visualising and analysing fluid flow are complex and expensive, using high power pulsed lasers and specialised cameras or numerous sensors to resolve the velocity and direction of particles in the fluid flow being analysed. By comparison PSV is based on one cheap, low-power continuous wave laser, coupled with an acousto-optic modulator, lenses, a cheap ordinary digital camera and fast real-time analysis software, and can be built for less than a tenth of the cost of the currently available systems.

[0148] PSV can provide instantaneous real time vectorisation of any size of flow field, ranging in size from supersonic aeroplanes to inside blood vessels. PSV's small size, light weight and durability will allow it potentially to be incorporated into flying or floating objects, or vehicles to provide real-time vectorised imaging of the flow near the critical parts of those moving objects. As well as traditional in-lab applications for the instrument, one application for a PSV unit could be installed on vehicles in use, for example on a flying aeroplane or during a test drive of a racing car, to provide the designers of the vehicle with instantaneous information about the lift and drag forces acting on the vehicle as a whole or on separate details of it. Another is in the observation of heart valve efficiency and blood flow. In this regard, the optics for radiation source and sensor could include waveguides, such as fibre optic arrangements, allowing for the fluid flow to be in a confined or generally inaccessible space (e.g. small ventilation holes, places behind moving wheels, blood vessels etc.).

[0149] Accordingly, as well as PSV being substituted for PIV, there are many situations in which PrV cannot be applied (in living organisms, real-time measurements, objects too large, etc.) PSV is then not a cheaper substitute of an existing technology, but a novel technique that opens new opportunities for the researchers and engineers.

[0150] Throughout this specification and claims which follow, unless the context requires otherwise, the word "comprise", and variations such as "comprises" or "comprising", will be understood to imply the inclusion of a stated integer or group of integers or steps but not the exclusion of any other integer or group of integers.

[0151] Persons skilled in the art will appreciate that numerous variations and modifications will become apparent. All such variations and modifications which become apparent to persons skilled in the art, should be considered to fall within the spirit and scope that the invention broadly appearing before described.

Claims

THE CLAIMS DEFINING THE INVENTION ARE AS FOLLOWS:

1) Apparatus for performing particle streak velocimetry, the apparatus including:

a) a radiation source that generates electromagnetic radiation for exposing particles in a fluid flow;

b) a sensor that senses radiation from the particles; and,

c) a control system coupled to the sensor and the radiation source, wherein the control system:

i) causes the radiation source to generate radiation having an intensity that varies continuously over an exposure time period;

ii) receives signals from the sensor, the signals being indicative of radiation from the particles during the exposure time period; and,

iii) generates an indicator indicative of at least a direction of travel of particles in the fluid flow.

2) Apparatus according to claim 1, wherein the intensity varies at least one of:

a) progressively over the exposure time period; and,

b) increases throughout the entire exposure time period.

3) Apparatus according to claim 1 or claim 2, wherein the control system includes at least one electronic processing device that processes the signals from the sensor.

4) Apparatus according to claim 3, wherein the electronic processing device generates an image showing streaks corresponding to the moving particles, and wherein the intensity of the streak varies in accordance with the varying intensity to thereby indicate a direction of movement.

5) Apparatus according to claim 3 or claim 4, wherein the electronic processing device generates a vector map indicative of the flow of particles in the fluid, the vector map including at least one vector indicative of a magnitude and direction of movement of at least one particle within the fluid flow.

6) Apparatus according to claim 5, wherein the electronic processing device derives the vector map from an image.

7) Apparatus according to claim 5 or claim 6, wherein the electronic processing device generates the vector map by:

a) identifying intensity peaks within the signals received from the sensor;

b) identifying ridges using the intensity peaks; and, c) analysing at least one ridge to determine a vector indicative of at least one of a direction and magnitude of particle movement.

8) Apparatus according to claim 7, wherein the electronic processing device:

a) generates an image intensity map;

b) removes background noise from the intensity map; and,

c) identifies intensity peaks in the intensity map.

9) Apparatus according to claim 8, wherein the electronic processing device removes the background noise using at least one of:

a) an intensity threshold; and,

b) a frequency-domain filter.

10) Apparatus according to claim 8 or claim 9, wherein the electronic processing device: a) enhances the intensity peaks using a filter;

b) performs edge detection to identify ridge edges; and,

c) generates a ridge map using the ridge edges.

11) Apparatus according to claim 10, wherein the electronic processing device performs edge detection using at least one of:

a) a neighbour clustering algorithm; and,

b) an intensity gradient analysis.

12) Apparatus according to claim 10 or claim 11, wherein the electronic processing device: a) performs ridge isolation to identify individual ridges; and,

b) generates a vector for each ridge using the ridge intensity.

13) Apparatus according to claim 12, wherein the electronic processing device:

a) identifies a maximum intensity for each ridge, the maximum intensity corresponding to a vector end; and,

b) identifies a ridge point furthest from the maximum intensity as a vector start.

14) Apparatus according to any one of the claims 7 to 13, wherein at least one electronic processing device:

a) determines changes in intensity along a vector;

b) compares the changes in intensity to the varying intensity of radiation generated by the radiation source; and,

c) uses a result of the comparison to determine at least one of:

i) components of movement at least one of: (1) perpendicular to a sheet of radiation used to expose the particles; and,

(2) parallel to the sheet; and,

ii) overlapping streaks;

iii) partly visible streaks; and,

iv) curved streaks.

15) Apparatus according to any one of the claims 1 to 14, wherein the radiation source includes a continuous intensity radiation source and a modulator for varying the intensity of the radiation.

16) Apparatus according to claim 15, wherein the radiation source is a continuous wave laser and the modulator is an acousto-optic modulator.

17) Apparatus according to claim 15 or claim 16, wherein the control system includes:

a) a pulse generator that generates an electrical control signal including at least one pulse having a defined pulse profile; and,

b) a driver coupled to the pulse generator that drives the acousto-optic modulator in accordance with the pulse profile of the control signal.

18) Apparatus according to any one of the claims 1 to 17, wherein the control system includes a trigger unit that generates a trigger signal to initiate the exposure time period.

19) Apparatus according to claim 18, wherein the trigger signal synchronises generation of the electromagnetic radiation of varying intensity and capturing of signals from the sensor.

20) Apparatus according to any one of the claims 1 to 19, wherein the sensor is a charge- coupled device sensor.

21) Apparatus according to any one of the claims 1 to 20, wherein the radiation source generates a substantially planar sheet of radiation, and wherein the image sensor is positioned substantially offset to and facing the sheet.

22) A method for performing particle streak velocimetry, the method including:

a) exposing particles in a fluid flow to electromagnetic radiation from a radiation source, the radiation having a continuously varying intensity over an exposure time period; b) using a sensor to sense radiation from the particles; and,

c) using signals received from the sensor during the exposure period to generate an indicator indicative of at least a direction of travel of particles in the fluid flow. 23) Apparatus for generating a vector map indicative of the flow of particles in a fluid, the apparatus including an electronic processing device that:

a) identifies intensity peaks within an image derived from radiation from particles in a fluid flow, the particles being exposed to electromagnetic radiation having a continuously varying intensity over an exposure time period;

b) identifies ridges using the intensity peaks; and,

c) analyses at least one ridge to determine a vector indicative of at least one of a direction and magnitude of movement of a respective particle.

24) Apparatus according to claim 23, wherein the electronic processing device:

a) generates an image intensity map;

b) removes background noise from the intensity map; and,

c) identifies intensity peaks in the intensity map.

25) Apparatus according to claim 24, wherein the electronic processing device removes the background noise using at least one of:

a) an intensity threshold; and,

b) a frequency-domain filter.

26) Apparatus according to claim 23 or claim 24, wherein the electronic processing device: a) enhances the intensity peaks using a filter;

b) performs edge detection to identify ridge edges; and,

c) generates a ridge map using the ridge edges.

27) Apparatus according to claim 26, wherein the electronic processing device performs edge detection using at least one of:

a) a neighbour clustering algorithm; and,

b) an intensity gradient analysis.

28) Apparatus according to claim 26 or claim 27, wherein the electronic processing device: a) performs ridge isolation to identify individual ridges; and,

b) generates a vector for each ridge using the ridge intensity.

29) Apparatus according to claim 28, wherein the electronic processing device:

a) identifies a maximum intensity for each ridge, the maximum intensity corresponding to a vector end; and,

b) identifies a ridge point furthest from the maximum intensity as a vector start. 30) Apparatus according to any one of the claims 23 to 29, wherein at least one electronic processing device:

a) determines changes in intensity along a vector;

b) compares the changes in intensity to the varying intensity of radiation generated by the radiation source; and,

c) uses result of the comparison to determine components of movement at least one of: i) perpendicular to a sheet of radiation used to expose the particles; and, ii) parallel to the sheet.

31) A method for generating a vector map indicative of the flow of particles in a fluid, the method including:

a) identifying intensity peaks within an image derived from radiation from particles in a fluid flow, the particles being exposed to electromagnetic radiation having a continuously varying intensity over an exposure time period;

b) identifying ridges using the intensity peaks; and,

c) analysing at least one ridge to determine a vector indicative of at least one of a direction and magnitude of movement of a respective particle.