WO2005043187A1

WO2005043187A1 - System and method for suppression of sea clutter in radar image

Info

Publication number: WO2005043187A1
Application number: PCT/GB2004/004463
Authority: WO
Inventors: Maurizio Bartozzi; Harold Limming; Simon Truss
Original assignee: Ledwood Mechanical Engineering Limited
Priority date: 2003-10-21
Filing date: 2004-10-21
Publication date: 2005-05-12
Also published as: GB0324620D0

Abstract

A method of sea clutter suppression in a radar image wherein a radar image is obtained (step 100) and for each pixel n of a total number N pixels (step 104), the brightness is determined (at step 106). Once this has been performed in respect of all N pixels another radar image is obtained and the process is repeated, until the radar, or at least the sea clutter suppression function, is switched off, at which point the process ends (step 116). In addition (after step 106), the average brightness change is calculated (at step 118) between pairs of subsequent updates for the same pixel n and the standard deviation of this average is also calculated (at step 120). The process then determines (at step 122) if the calculated standard deviation exceeds some predetermined threshold (say ± 1 σ) and, if so, this is recorded (at step 124) such that effectively the frequency with which new normalised brightness differences between pairs exceed the predetermined standard deviation threshold. At step 128 it is determined whether or not the recorded frequency exceeds same threshold value. If so, the pixel is identified as clutter-related and substracted from the image (step 134). At the same time, both the average brightness and standard deviation values are updated (at step 130) to take into account general condition variations.

Description

SYSTEM AND METHOD FOR SUPPRES SION OF SEA CLUTTER IN RADAR IMAGE

This invention relates to a radar system and, more particularly, but not necessarily exclusively, to a marine radar system , which may be either marine- or land-based.

Radar systems generally are used for detecting the presence or position or movement etc. of objects by sending out short radio waves which they reflect. In conventional radar systems, a marine radar scanner is used to transmit these radio waves, and a receiver receives the reflected waves. The received data is then processed and a view of the surroundings within the radar system range is displayed on a video screen. It is obviously important, particularly in the case of marine radar, to ensure that the clearest possible information is provided to a user about their surroundings and it is therefore desirable to ensure that the resultant video image is as clear and accurate as possible.

One problem encountered in many prior art systems is that of so-called "clutter" obscuring the image displayed on the screen and providing misleading information to the user about their surroundings. Strictly speaking, the term "sea clutter" refers to the imaging of the energy fraction associated with an incident radar pulse, which is backscattered by the sea surface towards the radar receiver. Without intervention, this results in an image being displayed which appears to contain objects or targets which are not really present, and in any event obscures the true picture.

The fraction of backscattered energy depends on several factors, including the characteristics of the incident pulse, incidence angle and, of course, sea surface characteristics. Because the incidence angle can be considered to be a system calibration constant, this issue will not be considered in any further detail herein. Other "clutters" also exist, primarily "rain clutter" and "land clutter". Rain clutter is obviously of great importance in all types of radar imaging, whereas land clutter, although of importance in many types of radar imaging systems, is not thought to be of great importance in respect of marine radar imaging.

Thus, the incidence angle is an important factor in controlling the backscattered intensity, as, in the case of sea clutter, is the local orientation of the normal to the sea surface in the area from which a radar signal sample is being obtained. In the case of a perfectly calm sea, the normal is always coincident with that of the gravity field at the same coordinates and the case corresponds with that of the incident beam being the only variable. In the more commonly- occurring situation whereby the sea surface is disturbed by some degree of wave motion, the incident beam is most efficiently backscattered towards the receiver by the receiver-facing sides of single wave bodies (assuming, of course, that the receiver and transmitter are coincident, as they almost always are). The backscattering efficiency achieved in this case can easily equal that of metallic objects such as ships and even exceed that of smaller vessels, such as inflatable and rigid inflatable boats. Moreover, as the backscattered energy is normally not collimated in a tight beam, but rather diffusely scattered by the backscattering surface, the image of a wave-originated backscatter signal can be indistinguishable in size from that originating from a vessel.

The discussion on the relationships between different types of waves and their backscattered signals is complex, and will not be entered into in any great detail for the purpose of this specification. For the following, it is sufficient to consider that, at any given time, an area affected by wave motion will be characterised by the presence, of a large number of bright patches separated (partially or totally) by a comparatively darker background. The density, size and interconnectedness of these patches depend on the local sea and wind conditions. Thus, an area will retain its general characteristics for as long as the local sea and wind conditions remain unchanged, even if single wave-induced returns appear and wane with a frequency loosely related to the period and wavelength of the wave motion.

In prior art radar imaging systems, sea clutter has been traditionally dealt with by Constant False Alarm Rate (CFAR) circuits, which are part of the radar receiver hardware. When CFAR is activated, the neighbouring area of the sampled location is analysed (usually in range, i.e. in one dimension only) to identify and count the number of bright spike occurrences (caused by backscatter). The radar gain is then reduced and/or the signal threshold adjusted according to some function of this number, thereby effectively decreasing the number of bright patches represented on the radar screen to give a predetermined number of spikes (the false alarm rate). The main problem with this method (and with similar, more complex methods in practical use) is that no distinction is attempted between clutter and non-clutter patches, so that returns from vessels may be removed from the radar display, together with unwanted returns caused by clutter, if their intensities are similar.

We have now devised an arrangement which overcomes this problem. Thus, in accordance with a first aspect of the present invention, there is provided a radar system comprising a transmitter for transmitting a signal to an area or location to be analysed, a receiver for receiving a signal reflected back from said area or location, and means for processing said received signal to produce an image of said area or location, the system further comprising means for collecting over a period of time two or more respective pieces of data relating to a characteristic of one or more portions of said image, analysing changes in brightness between said two or more pieces of data in respect of a portion of said image to determine if said portion is clutter-related and, if so, removing said clutter-related portion from said image.

Also in accordance with the present invention, there is provided a method of removing unwanted clutter from a radar image, comprising the steps of collecting over a period of time two or more respective pieces of data relating to a characteristic of one or more portions of said image, analysing differences between said two or more pieces of data in respect of a portion of said image to determine if said portion is clutter-related and, if so, removing said clutter-related portion from said image.

The method and system of the first aspect of the present invention enable the distinction between target-related and clutter-related portions of a radar image by analysing changes over time of a characteristic of each portion of the image, i.e. the first aspect of the present invention provides a statistics-based approach to sea clutter analysis and suppression in radar systems.

It will be appreciated that the backscattered intensity at a single location on the radar image affected by wave motion will fluctuate in phase with the wave motion itself. In a preferred embodiment of the present invention, therefore, brightness statistics for each location (beneficially corresponding to each pixel) on the radar image are obtained over a period of time, say between 10 and 20 minutes, average brightness changes between subsequent pairs of subsequent updates for the same pixel are calculated, from which average the standard deviation is determined.

Beneficially, a stabilisation time is allowed to enable the statistics to stabilise (in a preferred embodiment, a fixed stabilisation time is allowed for the entire image), after which the frequency with which new normalised brightness differences between pairs exceed a predetermined standard deviation threshold (set, for example, at + 1 σ) can be evaluated. At the same time, it is preferred to keep updating both average and standard deviation values, to take into account general condition variations.

Both standard deviation and frequency values can be used to assess the nature of the radar return represented by a pixel. Standard deviation alone can be used to differentiate between "noisy" and "homogeneous" areas, thus providing a way to distinguish between cluttered and clutter-free sea surface. Arguably more importantly, a standard deviation-based classification of different local conditions is possible. For instance, well-separable standard deviation values can typically be associated with sea- and rain-clutter affected areas, with the former presenting generally wider brightness oscillation amplitude than the latter. Rain and sea clutter can be distinguished also in terms of frequency value, as rain-generated returns appear smaller and more densely grouped than sea wave ones, thus fluctuating at higher frequency.

The frequency parameter can also be used to distinguish between coastline and sea clutter. On a radar image, the position of a coastline is normally not entirely constant, but instead oscillates slightly around an average position. As a result, a pixel in correspondence of a coastline will appear to have wide and extremely frequent brightness changes. A similar effect appears at the edges of semi-static floating objects, such as buoys. In fact, the frequency with which changes exceed ± lσ, say, is so high that such pixels can be easily discerned from both rain and sea clutter, using the method of the first aspect of the present invention.

Thus, once clutter-related pixels have been identified, they can be subtracted off the radar image leaving real targets virtually unaffected. Interpretation of radar images can require a great deal of skill and experience in accordance with the presentation of a radar image in a traditional (prior) art display. For example, a quick glance at a radar picture cannot tell a user which radar returns correspond to moving targets, buoys or land. It requires time and attention to understand the situation being presented by a conventional radar display.

A further problem occurs within target tracking systems used to measure and predict the speed of radar targets that correspond to vessels. This problem occurs because vessels which pass or cross each other can appear to merge on the display, even though, in reality, they may cross or pass each other with several tens of metres separation. Likewise, vessels passing close to other targets, such as land or buoys, can also cause the respective radar returns to merge.

This can cause erroneous predictions to be made by trackers, when presented with this type of data. Radar targets are identified as some measure of connectedness of samples that exceed a decision threshold. When the measurement system only has amplitude of reflected energy as its measurement system (as in most conventional systems), multiple real world objects can appear on the radar display to have merged and are therefore identified by a tracking system as a single target. When this occurs with crossing or passing targets, the erroneous track predictions that can occur are target drop, target loss or track swap. When this occurs with vessels passing static objects, the track can be seduced resulting in track drops or tracks following edges of land or jetties/piers or stopping on buoys.

We have now devised an arrangement which overcomes these problems. Thus, in accordance with a second aspect of the present invention, there is provided a radar system comprising a transmitter for transmitting a signal to an area or location to be analysed, a receiver for receiving a signal reflected back from said area or location, and means for processing said received signal to produce an image of said area or location, the system further comprising means for collecting over a period of time image data relating to said area or location, and analysing said image data collected over said period of time to distinguish between one or more moving targets and/or one or more static targets in said image representing said area or location. Thus, by analysing the radar data over a period of time, an understanding can be gained of what is moving and what is not, and this can be used to inform the tracking algorithms used by a tracking system to reduce the likelihood of the problems outlined above. The time period over which to analyse the data is preferably selected to ensure a reasonable degree of confidence in the estimates being made. The operator can also be informed by means of the radar display in a way that can assist them in understanding the received data more quickly. The presentation of the additional information can, for example, take the form of a change in brightness of the radar returns. The resultant information can be provided to a tracking system or the like, or used to perform, for example, static target suppression or moving target enhancement, as required.

When attempting to match radar returns to tracked targets in the tracking algorithms referred to above, motion analysis can be used to assist so as to reduce the probability of returns that are not similar to the tracked target's velocity.

Thus, in summary, the system according to the second aspect of the present invention can be used to calculate a long-terms average of an image, and apply some form of image analysis to determine or "learn" which targets are moving and which are static, such that, for example, the static targets can be removed from the image or otherwise differentiated or separated from the moving targets in the image, and/or a user can be given appropriate information.

In the case of both the first and second aspects of the invention, the radar image is beneficially presented in the form of an 8-bit (256 level) or more video display, which improves the dynamic range of the system in comparison with prior art systems, significantly enhances the image processing ability of the systems and also enhances the user's ability to interpret the resultant image.

Thus, in accordance with a third aspect of the present invention, there is provided a radar system comprising a transmitter for transmitting a signal to an area or location to be analysed, a receiver for receiving a signal reflected back from said area or location, and means for processing said received signal to produce an image of said area or location, and n-bit video means for displaying said image, wherein n is 8 or more. The strength of a radar signal return from a target is known to decrease with range according to a fourth (or "quad") power law. As a result, returns from distant targets quickly become very small. In an attempt to counteract this problem and improve the sensitivity of the radar longer ranges in prior art systems, it is known to increase the energy of the transmitted pulse for long range operation by increasing the pulse length. A side effect of this is to decrease the range resolution, so for short ranges, where greater resolution but less energy is required, the pulse length is decreased in prior art methods.

In the case where there are two or more radar displays, all using data from the same radar scanner but set to display pictures of different ranges, the radar may not have the optimum pulse length for all of the display range settings, resulting in either an unacceptably low resolution on the shorter range displays, or a lack of sensitivity on the longer range displays.

We have now devised an arrangement which overcomes the problems outlined above. Thus, in accordance with a fourth aspect of the present invention, there is provided a radar system comprising a transmitter for periodically transmitting a signal to an area or location to be imaged, a receiver for receiving said signal reflected back from said area or location, and means for processing said received signal to produce an image of said area or location, the system further comprising means for selectively altering the pulse length of said signals being transmitted by said transmitter, and means for receiving and displaying an image of said area or location derived only from signals received in respect of transmitted signals of substantially the same pulse length.

Thus, in one embodiment of the fourth aspect of the invention, the transmitter (which is preferably a transceiver which is configured to rotate, so as to transmit and receive signals in a substantially azimuthal manner) is arranged to consecutively transmit radio signals having, say, three different respective pulse lengths. A temporary storage means is preferably provided (in a computer-based radar system) to temporarily store the received signal from each azimuth. Means are beneficially provided for selecting which of the received signals are to be used to create a single radar picture. Thus, it is not necessary to process each azimuth in the sequence in which the reflected signal is received, as in prior art systems. In order to facilitate this, it is possible (although not essential) to attach a data 'tag' to each azimuthal scan in order to define its angular position and the pulse length, together with any other data which may prove to be useful in characterising the scan.

Hence, by using a suitable pulse repetition frequency (which may be variable between azimuths) and switching between a multiplicity of pulse lengths (say two or three, and not necessarily switched sequentially) and then selecting the correct azimuths (i.e. those having the same pulse length) from the memory, it is possible to build up a plurality of radar pictures, each displaying an image using a different pulse length. Thus, in accordance with the fourth aspect of the invention, all displays in a radar system (which, by way of example, may comprise a computer network) may choose to display a picture at any available range setting using the optimum pulse length for that range, irrespective of the requirements of other displays or users of the system.

Received radio energy transmitted by a rotating radar transceiver needs to be presented to the user in such a way as to represent a plan view of the cartesian world from which it was captured. In prior art radar systems, this transformation and representation of image data has required complex software, a very high processing capacity and a relatively large amount of time to be achieved.

We have now devised an arrangement which overcomes the problems outlined above. Thus, in accordance with a fifth aspect of the present invention, there is provided a radar system comprising means for receiving data representative of radio signals reflected back from an area or location to be imaged, and means for texture mapping the resultant data onto a three- dimensional polygon for display.

The resultant image is of better quality than that achieved in prior art systems, the processing is performed much more quickly, and the processing capacity requirement is significantly reduced, thereby freeing up the CPU so it can perform other duties.

These and other aspects of the present invention will be apparent from and elucidated with reference to the embodiments described hereinafter. Thus, embodiments of the present invention will now be described by way of examples only with reference to the accompanying drawings, in which:

Figure 1 is a flow diagram illustrating a method of sea clutter suppression according to an exemplary embodiment of the first aspect of the present invention;

Figure 2 is a schematic diagram illustrating the basic configuration and manner of operation of a radar system according to an exemplary embodiment of the fourth aspect of the present invention; and

Figure 3 is a schematic diagram illustrating the concept of texture mapping used in accordance with an exemplary embodiment of the fifth aspect of the present invention.

Referring to Figure 1 of the drawings, there is provided a flow diagram illustrating a method of sea clutter suppression in a radar image according to an exemplary embodiment of the present invention. At the start (step 100), a radar image is obtained. For each pixel n of a total number N pixels (step 104), the brightness is determined at step 106. The process then determines, at step 108 whether this has been performed in respect of all N pixels. If not, n is incremented by 1 (step 110) and the process returns to step 106. If so, another radar image is obtained and the process is repeated, until the radar, or at least the sea clutter suppression function, is switched off, at which point the process ends (step 116).

In addition, after step 106, the average brightness change is calculated at step 118 between pairs of subsequent updates for the same pixel n and the standard deviation of this average is also calculated (at step 120). At step 122, the process determines if the calculated standard deviation exceeds some predetermined threshold (say + 1 σ) and, if so, this is recorded (at step 124) such that effectively the frequency with which new normalised brightness differences between pairs exceed the predetermined standard deviation threshold. In any event, the process then increments n by 1 (at step 126) and returns to step 118.

At step 128 it is determined whether or not the recorded frequency exceeds same threshold value. If so, the pixel is identified as clutter-related and subtracted from the image (step 134). At the same time, both the average brightness and standard deviation values are updated (at step 130) to take into account general condition variations. In other words, a pixel is defined as clutter-related if the frequency with which its brightness change exceeds the brightness charge STD threshold in turn exceeds a pre-determined threshold. This threshold may be empirically defined. Alternatively, it may be determined rather than pre-fixed.

As before, this continues until the process ends (at step 116).

Referring to Figure 2 of the drawings, a radar system according to an exemplary embodiment of the fourth aspect of the present invention comprises a rotating radar scanner 200 comprising a transceiver for transmitting and receiving radar signals in an azimuthal manner.

The scanner 200 is communicatively linked to a computer network comprising, in this case, three radar displays 201, 202, 203. An output line 204 is provided which carries reflected signals received via the scanner to a set of temporary memory locations 205. As illustrated in Figure 2, the system according to this exemplary embodiment of the fourth aspect of the present invention is arranged to sequentially switch the pulse length of the signal transmitted by the scanner between Pulse length 1, Pulse length 2 and Pulse length 3. Thus, the pulse length of azimuth 1 might be Pulse length 1, and the respective received signal may be stored in a first memory location. The pulse length of the next azimuth, n + 1 , might be Pulse length 2, and the respective received signal may be stored in a second memory location. The pulse length of the next azimuth, n + 2, might be Pulse length 3, and the respective received signal may be stored in a third memory location. The pulse length of the next azimuth, n + 4, might once again be Pulse length 1, and the respective received signal may be stored in a fourth memory location, and so on.

Each of the radar displays 201, 202, 203 may have a different range setting, corresponding to each of the above-mentioned pulse lengths respectively. Thus, the range setting of display 201 may correspond to Pulse length 1 , the range setting of display 202 may correspond to Pulse length 2, and the range setting of display 203 may correspond Pulse length 3. Thus, all of the image signals received in response to the transmitted signal at Pulse length 1 is provided to display 202, where the radar image derived from all of the Pulse length 1 signals is displayed. Similarly, displays 202 and 203 display the radar images derived from the radar scans of Pulse lengths 2 and 3 respectively.

Referring now to Figure 3 of the drawings, and as stated above, received radio energy transmitted by a rotating radar transceiver needs to be presented to a user in such a way as to represent a plan view of the cartesian world from which it was captured. In accordance with an exemplary embodiment of the fifth aspect of the present invention, the stored image used for presenting the picture to the user is produced by either a forward or reverse scan conversion process. A PC graphics card or the like may be used to represent three-dimensional polygonal data on screen, by performing all of the transformations of the view matrix, lighting, colour, texturing and fog, etc. in hardware with sufficient speed to present many thousand of polygons per second. Texture mapping is the process of mapping part of a rectangular grid of numbers to a three-dimensional polygon (as illustrated schematically in Figure 3 of the drawings) and projecting it onto a two-dimensional surface which, for example, can then be stored in a frame buffer. The numbers stored in the rectangular grid are often a representation of colour information that form images used to paint the texture with.

By using a three-dimensional graphics programming API in a radar system according to an exemplary embodiment of the fifth aspect of the present invention, to describe the mapping of radar measurements to scan converted radar image, the graphics card can produce the scan converted radar image. Describing the radar image as a tiling of polygons with their associated texture mapping, the graphics hardware can efficiently render the radar data through a sequence of vector transformations and texture mapping. This frees up the CPU so that it can perform other duties.

Embodiments of the present invention have been described above by way of examples only, and it will be apparent to a person skilled in the art that modifications and variations can be made to the described embodiments without departing from the scope of the invention as defined by the appended claims.

Claims

CLAIMS:

1. A radar system comprising a transmitter for transmitting a signal to an area or location to be analysed, a receiver for receiving a signal reflected back from said area or location, and means for processing said received signal to produce an image of said area or location, the system further comprising means for collecting, over a period of time, two or more respective pieces of data relating to a characteristic of one or more portions of said image, analysing changes in brightness between said two or more pieces of data in respect of a portion of said image to determine if said portion is clutter-related and, if so, removing said clutter-related portion from said image.

2. A system according to claim 1 , wherein brightness statistics for each location on the radar image are obtained over said period of time, average brightness changes between subsequent pairs of subsequent updates for the same location are calculated, from which average the standard deviation is determined.

3. A system according to claim 2, wherein each location corresponds to a respective picture element.

4. A system according to claim 2 or claim 3, wherein a stabilisation time is allowed to enable the brightness statistics to stabilise.

5. A system according to claim 4, wherein a fixed stabilisation time is allowed for the entire image after which the frequency with which new normalised brightness differences between pairs of locations exceed a predetermined standard deviation threshold are evaluated.

6. A system according to claim 5 , wherein average and/or standard deviation values are also updated after said fixed stabilisation time.

7. A method of removing unwanted clutter from a radar image, comprising the steps of collecting over a period of time two or more respective pieces of data relating to a characteristic of one or more portions of said image, analysing differences between said two or more pieces of data in respect of a portion of said image to determine if said portion is clutter-related and, if so, removing said clutter-related portion from said image.

8. A radar system comprising a transmitter for transmitting a signal to an area or location to be analysed, a receiver for receiving a signal reflected back from said area or location, and means for processing said received signal to produce an image of said area or location, the system further comprising means for collecting over a period of time image data relating to said area or location, and analysing said image data collected over said period of time to distinguish between one or more moving targets and/or one or more static targets in said image representing said area or location.

9. A system according to any one of claims 1 to 6, wherein said radar image is presented in the form of an 8-bit (256 level) or more video display.

10. A radar system comprising a transmitter for transmitting a signal to an area or location to be analysed, a receiver for receiving a signal reflected back from said area or location, and means for processing said received signal to produce an image of said area or location, and n-bit video means for displaying said image, wherein n is 8 or more.

11. A radar system comprising a transmitter for periodically transmitting a signal to an area or location to be imaged, a receiver for receiving said signal reflected back from said area or location, and means for processing said received signal to produce an image of said area or location, the system further comprising means for selectively altering the pulse length of said signals being transmitted by said transmitter, and means for receiving and displaying an image of said area or location derived only from signals received in respect of transmitted signals of substantially the same pulse length.

12. A system according to claim 11 , wherein the transmitter is arranged to consecutively transmit radio signals having different respective pulse lengths.

13. A system according to claim 11 or claim 12, wherein a temporary storage means is provided to temporarily store the received signal from each signal transmission.

14. A system according to any one of claims 11 to 13, wherein means are provided for selected which of the received signals are to be used to create a single radar picture.