Radar: Radar-video extractor for ship detection (1965 – 1979)


The development of the radar-video extractor for the detection of ships

Radar is widely used for detecting objects to control traffic. In the beginning, this was only done by looking at a radar screen. That did happen with an ‘expert eye’ because beyond important objects (aircraft, ships) there is much more to see. Everything that reflects radar waves is made visible and disturbs the observer’s image. This includes reflections of the water surface, buildings, rain clouds, and so on. Besides, there are interferences with radio and radar sources, ghost echoes and noise, in short, a multitude of information so that only a trained operator could separate the wheat from the chaff. With such a complex image it is difficult to apply automatic signal processing. For that purpose, it is necessary that the echoes of interest, “the targets”, can electronically and automatically be separated from the rest of the radar signal. A radar-video extractor has been developed for this purpose, a device that detects the targets from the total radar signal based on many parameters and thereby makes the position, and sometimes also the dimensions, of the targets electronically available for further processing such as automatic target tracking.

This technique was first developed for the detection of aircraft. That is the easiest. The echo of an aeroplane is almost always the same and therefore reasonably good to sketch. The echo is defined by the length of the transmit pulse of the radar, the width of the beam of the radar antenna, and the rotational speed thereof. The dimensions of the aircraft play a negligible role here. We, therefore, speak of ‘point targets’. The first known extractor for detecting aircraft was developed in the mid-1960s and used the principle of the ‘moving window’. Within a time window relating to the distance and rotation of the radar, the number of signal returns exceeding set thresholds for the radar signals was counted. The resulting count determines whether there is a target in the window. This went fairly inaccurately because also with a certain amount of clutter (false echoes) and noise, a target is detected. As a result, many ‘false alarms’ were reported.

The development of radar-video extraction at TNO

In the mid-60s, J. (Joop) P. Kunz of the Physics Laboratory NDRO (Netherlands Defence Research Organisation) started to realise his idea of an extractor of the type ‘detector-accumulator’. The radar signal of several transmit pulses is integrated causing reflections of targets to be amplified. Noise and clutter are toned down. This method proved to be superior to the ‘moving window’ extractor. The main disadvantage remained: very useful for detecting point targets (aircraft) but useless for targets whose dimensions are large as compared to the parameters of the radar (transmit pulse length and antenna beam). The detection of ships was therefore still not feasible.

At the beginning of 1970, a new method was being sought to automatically detect ships. This is motivated by the automation of sea-warning radars at the Royal Netherlands Navy and by the arrival of Side Looking Airborne Radar (SLAR) used by Rijkswaterstaat for observations of ships by aircraft. The idea was born to develop a radar video extractor that could be dimensioned so that the output of all types of radars could be processed. This project was managed by ing W. (Wim) FM van der Heijden and Mr E. (Eddy) L. Intres. A combination of the moving window extractor and the detector accumulator extractor was considered, bypassing the drawbacks of both methods.
Because the specifications of a sea-warning radar (long-range, long transmission pulse with low repetition rate from a slow rotating antenna) for the Royal Netherlands Navy, a SLAR (medium-range, medium transmit pulse with high repetition frequency from a linear antenna) and a port radar (short-range, short transmitting pulse with a high repetition rate from a rapidly rotating antenna), it was decided to make the extractor to be developed as universal as possible. All parameters for detecting ships should be fully adjustable, as should the adjustments to the radar and the antenna to which the device should be connected. Of course, the extractor had also to be able to detect aeroplanes from an air warning radar. And that’s how the “Automatic measuring extractor with adjustable parameters” was developed.

In parallel, a start was made with the renewal of the shore radar system for the port of Rotterdam. The old radar system was only suitable for visual observations. The port needs automatic processing of radar targets and automatic link of the identity of ships and their position. Therefore, radar-video extraction is the link between the analogue radar and the digital processing of data. For that reason, the Port Authority of Rotterdam invited market parties for demonstration projects. One project was the automatic processing of radar data using extraction technology. The demonstration provided by a potential supplier, however, was disappointing. The Pilotage, at that moment still part of the Royal Netherlands Navy and future co-user of the new shore radar system, raised a large number of questions about the outcome of that demonstration without immediately being able why it did not function properly. To this end, the Pilotage Service approached the experts from the Physics Laboratory NDRO (Netherlands Defence Research Organisation). Research showed that the demonstration used a radar extractor of the ‘moving window’ type that is unsuitable for the detection of ships at a relatively short distance. The devastating TNO report put an end to the demonstration.
This caused the Walradar renewal project to get stuck in an impasse because there were no alternatives available. TNO, however, indicated that a new idea for radar extraction was in development specifically developed for the detection of ships.

Automatic measuring extractor with adjustable parameters

The extractor under development was not a prototype for a particular application. The ability to adapt all parameters to the radar, the targets to be detected and the environmental situation allowed this extractor to be used for measurements to determine the final set of parameters and specifications for an operational system. The extractor was realised in two phases. The first development phase was deployed to detect ships at sea from an aircraft with the SLAR, which was then used by the Physics Laboratory TNO in collaboration with Rijkswaterstaat. Because the antenna was fixed to the aircraft, the processing of angular information had not yet been implemented. Although a few teething problems were detected, ships could be well detected.

The renewal of the Rotterdam shore radar system led to the start of the second phase. The Physics Laboratory TNO was commissioned to study the possibilities of automatic radar extraction for the future shore radar system in collaboration with Hollandse Signaalapparaten BV (HSA), Hengelo. The experimental extractor was made suitable for fast-rotating, short-range search radars with a high pulse repetition frequency. Because the electronics were not fast enough at the time, a decompression technique had to be developed to achieve sufficient resolution for a good determination of the position, dimensions and location of large ships at a short distance. To prevent the overloading of a connected computer system for the processing of the targets, work was also done to add a masking technique to the extractor. All radar information outside the shipping lanes was filtered out so that only information about ships had to be processed. Finally, the detection algorithms were improved so that radar echoes that split or grow correctly could be better detected. To perform analysis on certain targets, a light pen was added that made it possible to track and analyse a selected target manually. That saved a lot of calculation work after a study. After all these adjustments, the extractor was ready as a measuring device. In 1976, a series of tests with the extractor was carried out in the Rotterdam port area. The study resulted in a series of recommendations and parameter specifications for an operational radar video extractor in a port area.

Tests for the shore radar system at the Sluisjesdijk in Rotterdam. On the table is the measuring extractor, above that the video recorder to record the radar signals.
Tests for the shore radar system at the Sluisjesdijk in Rotterdam. On the table is the measuring extractor, and above that is the video recorder to record the radar signals.

As a result of these measurements, HSA purchased the principle of this extraction technique developed by TNO. HSA paid for the exclusive rights of using this technology NLG 40,000, being the hardware costs incurred for the construction of the measuring extractor. TNO did retain the right to use the measuring extractor for other projects. HSA applied for and obtained a patent on this extraction principle in 1979 in which the TNO employees van der Heijden (author) and Intres are named inventors. The new shore radar system for Rotterdam was then built by HSA and supplied with this extraction principle. It was fully operational in 1987.

Of course, something has changed afterwards. When this measuring extractor was developed, no computer was able to do the required real-time calculations quickly enough to extract ship targets from a radar signal. That is why a specialised device was developed that was the first to be able to automatically detect a ship from a radar signal and to make the information suitable for automatic processing with the then-available technology and on the edge of the technical possibilities. Today one does that with a standard PC. However, TNO’s success lies in the fact that the underlying method is still in use worldwide for port radar systems produced by various manufacturers.

The described measuring extractor was used afterwards in various projects, not only to determine the parameters for an extraction system to be built but also to carry out analysis on board navy vessels to improve operational availability. This was done with professional video recorders that were adapted by TNO to record the radar-video signal. The adjustment of the synchronisation of the video heads was specifically important to be able to playback the radar signals in a realistic way. The recorded signal could then be repeatedly played back and presented to the measuring extractor so that with the parameters set differently, an optimal radar presentation (both for display and processing) could be determined. Some Navy radars have more radar signals available with different RF receivers. All these signals were recorded for analysis on more video recorders and offered to the extractor to determine the optimal target detection.


This page is based on an excerpt made by Ing. W.(Wim) F.M. van der Heijden