Skip to end of metadata
Go to start of metadata

You are viewing an old version of this page. View the current version.

Compare with Current View Page History

Version 1 Next »

Introduction

This page details the settings available for the tracker. These settings allow the tracker for each radar to be optimised for each particular site or target type.

Contents

Advanced Settings

Processes

Use STC: Select this to enable Sensitivity Time Control to reduce clutter from nearby static objects. STC helps to reduce clutter by selectively attenuating the radar signal power, within a specified range, to allow for nearby interfering factors, such as buildings, gantries, lakes etc.

Show STC Range: Select to show the STC Range.

Use FTC: Select this for the tracker to use a Fast Time Constant high-pass filter to help reduce the impact of rain on radar results. This is to be used in exceptional circumstances.

Use Clutter Map: Select this for the tracker to use the saved clutter map.

Use Clutter Guard: This adds a margin of error to established clutter in order to avoid false alarms being created from real-time fluctuations in signal return.

Use Azimuth Blanking: Select this to remove all raw radar data from the processing chain behind large returns close to the radar, to avoid unnecessary false alarms. If a track is a certain size at a certain range, we reduce the target shadow to limit false recording by removing signals in the shadow of large targets. Under normal circumstances, we would want to look for multiple targets in the same azimuth, however, when we have large targets very close to the radar, this prevents us detecting anything behind them. In this scenario it is safer to dynamically suppress potential targets, but only for as long as it takes the large target to move out of range.

Use Static Target Detection: Select this to protect long-stationary targets from being absorbed into the clutter map.

Thresholding

Threshold Delta (pwr): Defines the threshold above which the signal will be included for track processing. Signal levels below this threshold will be discarded. Setting a lower threshold value provides greater detection capability but can correspondingly increase the level of false alarms.

Threshold Window (m): The moving window size over which to calculate the dynamic threshold. The noise is averaged over this window size and the threshold delta is applied to this averaged value to check for significant power returns.

Threshold Output Value: Defines the value that will be sent through to the tracker when a radar input signal exceeds the Threshold Delta value and how the input signal from that radar should be conditioned before being passed on to the next stage.

Input Value: Passes on the radar signal unchanged, so the total amplitude of each signal wave is recorded.

Level Above Threshold: Subtracts the Threshold value from the radar signal to isolate only the signal that exceeds the threshold.

Full Scale: Converts any signal that exceeds the Threshold essentially into a binary signal (i.e. either zero or the maximum value).

SD Factor: A factor used to clean up the process data by using the standard deviation of the local threshold data.

STC

Range: Defines the distance up to which the attenuation specified in the Power Factor field should be applied.

Start Range: Defines the distance from the radar to the point at which the Power Factor increase is triggered. In the graph below, it is hatched in green.

Power Factor (%): Determines the attenuation that should be applied to operation within the distance to local objects that are specified in the STC Range field.

Once the Use STC option is selected, the tracker will apply the chosen Power Factor after the start range, and scale up (in linear fashion) to 100% of the stated STC Range. As a result, the effect of nearby objects is reduced.

This is especially useful when the radar is near water, to reduce signal reflection.

FTC

FTC Range (m): Defines the width of the high-pass filter.

FTC Level (0-1): Defines the local average that must be subtracted from the input signal to achieve the high-pass filter.

Clutter Map

Clutter Map Time (%): This value dictates how much of the received radar data contributes towards maintaining an accurate Clutter Map and movement analysis. It is measured as a data percentage per revolution of the radar, e.g. for a value of 0.85, this means 15% of the data from the scan is used to update the clutter (background radar picture). Therefore 85% of the data is analysed for movement. The higher the number, the more the data is analysed for movement, and the more sensitive it is to movement. Recommended setting is between 0.8 and 0.95.

Clutter Startup Delay (s): This value determines clutter absorption period.

Clutter Slow Decay: The Clutter Slow Decay effects how clutter is removed from the clutter map. It works by adding to the Time Constant (please refer to Tracker) being used in the clutter map. We use a default Time Constant of 0.90 in the standard tracking clutter map, so the slow decay is a % increase of the remaining 0.10 from the Time Constant. For example: a 50% Slow Decay is equal to 50% of the 0.10 that is not used by the clutter channel (so it would be 0.05). This is then added to the Time Constant when clutter is being removed, causing the removal of clutter to be slower (essentially being removed at a Time Constant of 0.95 compared to the previous 0.90). This allows us to hold onto the clutter data for longer. An example of this useful feature can be seen in ClearWay™ during situations where queues are in progress. When the queue drops, we often get false tracks where targets, such as trees, become visible to the radar again. By holding the clutter data for longer, there is more time before the clutter value will drop below the threshold, causing an alarm when the clutter jumps back up, once the queue has cleared.

Use Clutter File as Floor: Select this for the tracker will use the saved clutter map as the base template of static objects (to be ignored) within sight of the radar sensor.

Hold Azimuth Multiplier (width) and Hold Range Multiplier (depth): These parameters are used to prevent tracks that become stationary for a period of time from being absorbed into the clutter map. To do this the clutter map process holds open a hole in the clutter map to prevent the clutter building up where the track is. These two parameters define the size of the ‘hole’ as a factor of the reported track size. A value of 1 for both range and azimuth holds an area in range and azimuth that is the same size as the track. If they are both set to 2 then the area would be twice the size of track and so on. Using a value slightly larger than 1 allows for any variation in the reported size of track. It also allows for very small movement when the track is stationary. If both values are set to zero then track holding is disabled.

Clutter Guard

Threshold (pwr):

Noise Min Value (pwr):

Noise Gain:

Signal Gain:

Signal Offset:

Signal Speed:

The Clutter Guard is to be used with caution because it will reduce detection performance in and around clutter.

Azimuth Blanking

Threshold (pwr): Defines the power of the signal strength.

Maximum Bin: The maximum number of bins (individual cells measured by radars) in which to look for a target above the Threshold.

Bin Tolerance: As the radar azimuth can only see the front of a target, this setting is used to approximate how deep the target is.

Show Blanking: Typically, this setting is only used whilst configuring and testing the Azimuth Blanking, as the radar display will not show the data behind a selected target. On the PPI this would display as a blank segment behind a target when viewing raw data source.

Radar resolution is different for each model, therefore, to calculate Maximum Bins, you need to know the range resolution of the radar.

Azimuth Blanking is a destructive process as it removes data behind the target, whereas Shadow Suppression suppresses the track creation in the shadow of a target but does not remove data.

Static Target Detection

Threshold (%): The minimum % that the live score should be above the base line score.

Base Line Samples: The number of samples to use to calculate the base line score.

Live Samples: The number of samples to use when calculating the live score to compare with the base line.

Channels

Channel

There will be as many channels as you have carriageways. Each carriageway's channel can be individually configured.

Plot Extraction

Plots are grouped cells of the radar's detection range that have been recorded as containing an object.

Min Amplitude (pwr): Determines the minimum amplitude of a plot to be considered for extraction.

Min Plot Range (m): Determines the minimum size in range of a plot to be considered for extraction.

Min Range Span (bins): Determines the minimum number of consecutive range samples that must be in a plot before it is considered for extraction.

Min Blob Weight (cells): Determines the minimum number of cells (where a cell is a single power reading in either range or azimuth) that must be in a plot before it is considered for extraction.

Plot Merging

Once plots are extracted (using the ‘minimum’ settings mentioned above), this option determines whether plots that are in close proximity to each other should be merged into single, larger plots using the method selected in the Mode setting. E.g. a large articulated truck might return several plots (cab, trailer, etc.) but the operator needs to see this as a single object. There is no way for the radar to know if two separate plots are part of the same target or if each plot is a different target, however, plot merging works on the basis that if two plots are “close” enough then we consider them to be part of the same target, so we merge them together into a single plot / target. To achieve this, plots need to be intelligently merged together without combining smaller individual vehicles. The different modes decide how you want to merge and also how “close” they need to be before we merge.

There are three modes:

Azimuth Merging

This mode looks at the gap between plots in azimuth and also the distance they are apart.

This mode is good for targets which are moving perpendicular to the radar and are further away from the radar. Note that this uses Merge Azimuth and Merge Range. But like all azimuth measurements, these become less accurate with range because of the beam width spread.

Distance Merging

This mode looks at the distance between the centre of the plots and ignores the shape of each plot. This is quite a good approach for targets which have fragmented in multiple plots. In some situations, especially close to the radar, this approach is better than Shape Distance.

This only uses the Merge Range setting.

Shape Distance Merging

This mode looks at the distance between the edge of each plot based on the shape of each plot. So it will only merge plots which are within the specified distance from the edge of each plot. This mode is good for targets where the shape of the plot is clear.

This only uses the Merge Range setting.

Merge Range (m): How far to look for the next plot to merge in range. Consider this a 'merge window' - the value represents the length of the window along the range that looks for plots to merge.

Merge Azimuth (°): How far to look for the next plot to merge in azimuth. Consider this a 'merge window' - the value represents the width of the window along the azimuths that looks for plots to merge.

We recommend that Distance Merging and Shape Distance Merging are the most effective for traffic.

Users should be careful about correct usage of plot merging. Two extreme cases are possible:

  • A large vehicle may generate many large plots (as the radar beam sweeps the vehicle, it could hit it in several places). With plot merging turned off (or misconfigured) these plots would appear to be unconnected to each other, and would be more likely to be classified as several targets.

  • At the other end of the spectrum, several targets which are close to each other could be merged into a single track if plot merging is turned on and set too aggressively. For example a group of people could be merged together and this would have the potential to be classified as a vehicle.

Tracking

Initial Delay (s): A timeout period after start up, during which will not generate any tracks to allow a grace period whilst the tracker establishes the background clutter, etc.

Clutter Track Age Limit (days): The maximum period of time that Clutter Map data can be retained and used.

Filter Mode: This setting controls the tolerances used by the tracker for predicting movement.

Fixed Gain: This is the default setting. Fixed Gain is the most reliable setting for predicting movement, but it is less tolerant of tracks that are manoeuvring in an unexpected manner.

Dynamic: The Dynamic setting will use a combination of tolerances depending on what the track is doing. It will try and use Fixed Gain for targets moving in a predictable fashion, yet, when required, will be more tolerant of targets making unexpected movements; for example, tracking a fast-moving vehicle all the way around a bend.

Tracking Areas

Area: Select one of the Areas from the list located in the Configuration Tree.

Parameters: Select the Tracking Parameter from the Configuration Tree.

Proximity Suppression

Track Margin (m): The radius of the Proximity Suppression Area.

Wake Suppression

This is typically used to suppress the generation of unwanted targets caused by the wake behind a boat. This feature calculates a suppression zone based on an angle and distance behind the boat. There are two options for calculating the distance, it can either be a fixed distance or can be based on time. Note that the time option includes an expansion factor which adjusts the suppression zone based on the size of the boat.

Wake Angle (°): θ , Theta. The angle of the wake spreading from the boat.

Wake Distance (m): x, the length of the wake.

Wake Time (s): Stipulate a time, t, to calculate the size of the suppression zone. You can use this alongside or instead of Wake Distance. If both Wake Time and Wake Distance are configured, then the distance used for suppression will be the maximum of the two values.

Wake Range (m): The maximum Wake Distance.

When using Wake Time an expansion factor is included in the calculation that automatically adjusts the size of the wake suppression area based on the size of the target. The calculation used is as follows:

Wake Time * radar update rate * expansion factor

The expansion factor is as follows: 1 = normal size boat, 2 = medium sized boat, 4 = large boat

Because the Wake Time is based on the radar update rate it means that a value of 1 does not necessarily mean 1 second. For example if you are using a 2Hz radar than a wake time of 2 would actually be 1 second. If using a 1Hz radar then a value of 2 would be 2 seconds.

Shadow Suppression

Shadow Angle (°): θ , Theta. The angle of the 'shadow' cast by the track obscuring the beam of the radar.

Shadow Distance (m): x, the length of the shadow.

Note that if you set the Shadow Angle to 0 (zero) then the angle will be automatically set based on the size the target. This is the default and recommended setting.

Proximity Time Suppression

Track Margin Time (s): Time (t) in seconds. Time x velocity (v) = distance ( x ) in metres. Velocity is speed, with the unit of m/s.

Tracking Areas

Area: Select one of the Areas from the list located in the Configuration Tree.

Parameter: Select the Tracking Parameter from the Configuration Tree.

Debris

Processes

Use Debris Channel: Select to enable the Debris Channel.

Clutter Map

Clutter Map Time (%): The system take a percentage of the incoming signal and adds it to the Clutter Map.

Clutter Start Up Delay (m): The time after start-up before the system processes further data. This helps with clutter occlusion and also helps speed up the start-up.

Clutter Slow Decay: The Clutter Slow Decay effects how clutter is removed from the clutter map. It works by adding to the Time Constant (please refer to Tracker) being used in the debris channel. We use a default Time Constant of 0.995 in the debris channel, so the slow decay is a % increase of the remaining 0.005 from the Time Constant. For example: a 50% Slow Decay is equal to 50% of the 0.005 that is not used by the debris channel (so it would be 0.0025). This is then added to the Time Constant when clutter is being removed, causing the removal of clutter to be slower (essentially being removed at a Time Constant of 0.9975 compared to the previous 0.995). This allows us to hold onto the clutter data for longer, which is useful in situations where queues are in progress. When the queue drops, we often get false tracks where targets, such as trees, become visible to the radar again. By holding the clutter data for longer, there is more time before the clutter value will drop below the threshold, causing an alarm when the clutter jumps back up, once the queue has cleared.

Use Clutter File as Floor: This is the minimum value the clutter can drop to.

The Clutter Floor recording needs to be as clean as possible

Thresholding

Threshold Delta: This threshold works in exactly the same way as for normal processing, but it only applies to the debris channel. The threshold output is always levels above, therefore there are the normal area dependent threshold settings in the Debris Default tracking parameters, however we would encourage you to only use this global threshold setting. If you increase this value, you will desensitise the detection, decreasing the value will increase sensitivity and therefore increase the likelihood of nuisance alarms.

Threshold Window: The moving window size over which to calculate the dynamic threshold. The noise is averaged over this window size and the threshold delta is applied to this averaged value to check for significant power returns.

Threshold Output Value: Defines the value that will be sent through to the tracker when a radar input signal exceeds the Threshold Delta value and how the input signal from that radar should be conditioned before being passed on to the next stage.

Input Value: Passes on the radar signal unchanged, so the total amplitude of each signal wave is recorded.

Level Above Threshold: Subtracts the Threshold Value from the radar signal to isolate only the signal that exceeds the Threshold.

Full Scale: Converts any signal that exceeds the Threshold essentially into a binary signal (i.e. either zero or the maximum value).

SD Factor: A factor used to clean up the process data by using the standard deviation of the local threshold data.

Scan to Scan Integration

Full Window Size (m) (hits): This provides the total number of scans used to examine a signal to establish whether we think it is debris. In the worst case this may lead to a 4 second delay in detecting debris, but this process is essential to detect small targets with any level of confidence. Increasing this window size will delay detection and potentially reduce false alarms. The disadvantage of delaying detection is that more of the target will become absorbed into the clutter map and therefore become harder to detect over time. This setting is linked to the Recent Window Size and the Hit Counts - they should all be considered when making changes to the configuration. The process will be looking for the Full Hit Count Threshold being exceeded within this window. For example, with a Full Window Size of 16 and a Full Hit Count Threshold of 10, we would expect our debris signal to appear above threshold 10/16 scans.

Recent Window Size (m) (hits): This window provides the total number of scan to analyse within the last n of the Full Window Size. This allows us to specify a level of confidence that indicates the same signal we saw at the beginning of the Full Window is still there by the end of the Full Window scan period. For example, if the Full Window Size is 16 and the Recent Window Size is 6 then this means use the last 6 scans (i.e. 11 - 16) of the full window as the recent window. Within this recent window we look for a further hit threshold using the Recent Hit Count Threshold. This has to be exceeded along with the Full Hit Count Threshold for the detection to be considered genuine debris. Increasing this value will typically indicate you expect more threshold hits over a larger proportion of the Full Window scan. Decreasing this value indicates you are focusing on hits at the very end of the scan period.

Full Hit Count: This is the total number of times we expect the threshold to be exceeded within the specified Full Window Size. If the Full Window Size is 16 and the Full Hit Count is 10, then we need to see a total of 10 hits in 16 scans to consider the target as debris.

Recent Hit Count: This is the total number of times we expect the threshold to be exceeded within the specified Recent Window Size. If the Recent Window Size is 6, and the Recent Hit Count is 3, then we need to see a total of 3 hits in last 6 scans to consider the target as debris.

Integration Threshold (pwr): This is the threshold for input video to be considered as significant for integration. If the video exceeds this threshold then it is integrated. When used with the Dynamic Threshold process configured, set this threshold to 0. The purpose of the threshold process is to decide if each video sample is “of interest” for further processing by comparing the video intensity to a threshold level. If the video sample exceeds the threshold then it is passed on. If it does not exceed the threshold the video sample is rejected. The thresholding process compares each video sample to a background threshold. The background threshold is calculated for each sample and reflects the local video distribution around the test sample.

Instant Threshold (pwr): Instant threshold allows cells of sufficient signal strength to bypass the SSI (scan to scan integration) process. So any cell with a strength above the Instant Threshold will immediately be made available for plot extraction.

Plot Extraction

Plots are grouped cells of the radar's detection range that have been recorded as containing an object.

Min. Amplitude (pwr): Determines the minimum amplitude of a plot to be considered for extraction.

Min. Plot Range (m): Determines the minimum size in range of a plot to be considered for extraction.

Min. Range Span (bins): Determines the minimum number of consecutive range samples that must be in a plot before it is considered for extraction.

Min. Blob Weight (cells): Determines the minimum number of cells (where a cell is a single power reading in either range or azimuth) that must be in a plot before it is considered for extraction.

Plot Merging

Merge Plots: Once plots are extracted (using the ‘minimum’ settings mentioned above), this option determines whether plots that are in close proximity to each other should be merged into larger plots using the method selected in the Mode setting.

Mode:

None: No plot merging.

Azimuth: Merges in azimuth. This option provides good performance for highway monitoring due to the fixed movement of vehicles in straight lines past the radar.

Distance: Merges in range distance (along the radius).

Shape Distance: Merges in azimuth and range. Attempts to draw a containing boundary around relevant plots. This is a good general purpose option and works well for most situations.

Merge Range (m): How far to look for the next plot to merge in range. Consider this a 'merge window' - the value represents the length of the window along the range that looks for plots to merge.

Merge Azimuth (°): How far to look for the next plot to merge in azimuth. Consider this a 'merge window' - the value represents the width of the window along the azimuths that looks for plots to merge.

Tracking

Initial Delay (s): A timeout period after start up, during which Witness™ will not generate any debris to allow a grace period whilst the radar establishes the background clutter, etc.

Tracking Areas

Area: Select one of the Areas from the list located in the Configuration Tree.

Parameter: Select the Tracking Parameter from the list located in the Configuration Tree.

Editing Multiple Radar

When highlighting more than one radar in the list within the radar dialog there will be the following selected values to edit:

Channel

Update Debris: Select this option to edit the channel settings marked with a * in the fields below:

Threshold Delta (pwr): Defines the threshold above which the signal will be included for track processing. Signal levels below this threshold will be discarded. Setting a lower threshold value provides greater detection capability but can correspondingly increase the level of false alarms.

Threshold Window (m): The moving window size over which to calculate the dynamic threshold. The noise is averaged over this window size and the threshold delta is applied to this averaged value to check for significant power returns.

SD Factor: A factor used to clean up the process data by using the standard deviation of the local threshold data.

Merge Range (m): How far to look for the next plot to merge in range. Consider this a 'merge window' - the value represents the length of the window along the range that looks for plots to merge.

Merge Azimuth (°): How far to look for the next plot to merge in azimuth. Consider this a 'merge window' - the value represents the width of the window along the azimuths that looks for plots to merge.

Full Window Size* (m) (hits): This provides the total number of scans used to examine a signal to establish whether we think it is debris. In the worst case this may lead to a 4 second delay in detecting debris, but this process is essential to detect small targets with any level of confidence. Increasing this window size will delay detection and potentially reduce false alarms. The disadvantage of delaying detection is that more of the target will become absorbed into the clutter map and therefore become harder to detect over time. This setting is linked to the Recent Window Size and the Hit Counts - they should all be considered when making changes to the configuration. The process will be looking for the Full Hit Count Threshold being exceeded within this window. For example, with a Full Window Size of 16 and a Full Hit Count Threshold of 10, we would expect our debris signal to appear above threshold 10/16 scans.

Recent Window Size* (m) (hits): This window provides the total number of scan to analyse within the last n of the Full Window Size. This allows us to specify a level of confidence that indicates the same signal we saw at the beginning of the Full Window is still there by the end of the Full Window scan period. For example, if the Full Window Size is 16 and the Recent Window Size is 6 then this means use the last 6 scans (i.e. 11 - 16) of the full window as the recent window. Within this recent window we look for a further hit threshold using the Recent Hit Count Threshold. This has to be exceeded along with the Full Hit Count Threshold for the detection to be considered genuine debris. Increasing this value will typically indicate you expect more threshold hits over a larger proportion of the Full Window scan. Decreasing this value indicates you are focusing on hits at the very end of the scan period.

Full Hit Count*: This is the total number of times we expect the threshold to be exceeded within the specified Full Window Size. If the Full Window Size is 16 and the Full Hit Count is 10, then we need to see a total of 10 hits in 16 scans to consider the target as debris.

Recent Hit Count*: This is the total number of times we expect the threshold to be exceeded within the specified Recent Window Size. If the Recent Window Size is 6, and the Recent Hit Count is 3, then we need to see a total of 3 hits in last 6 scans to consider the target as debris.

Recursive Average Gain*: The gain used when implementing a rolling average of the amplitude values of real cells seen during the SSI process. The gain controls how much of a new cell's amplitude is added to the rolling average.

Global

Use STC: Select this to enable Sensitivity Time Control to reduce clutter from nearby static objects. STC helps to reduce clutter by selectively attenuating the radar signal power, within a specified range, to allow for nearby interfering factors, such as buildings, gantries, lakes etc.

STC Range: Defines the distance up to which the attenuation specified in the Power Factor field should be applied.

Power Factor (%): Determines the attenuation that should be applied to operation within the distance to local objects that are specified in the STC Range field.

Distribute Tracks: Select this to display the tracks on the PPI carriageway. 

Layer: The layer of the PPI display on which the radar is active.

Group: Which radar group this radar belongs to e.g. Default Radar Group.

Hide: Select this to make the radar unselectable and translucent in appearance on the PPI. This allows you to simplify the clutter within the site map by concealing this item while leaving it working as usual. It can still be selected within the Configuration Tree - it will be temporarily revealed within the site map.

X and Y Offset Delta (m): These are X and Y offset deltas that are applied to the position of all the selected radar. For example, if you provided an X delta of +10m then all the selected radar would move east by 10m from their current location.

Related Information

  • No labels

Safety is everything.