CA3038596A1

CA3038596A1 - Detector unit and a method for detecting an optical detection signal

Info

Publication number: CA3038596A1
Application number: CA3038596A
Authority: CA
Inventors: Stefan Scherbarth; Jochen Lauber
Original assignee: Hensoldt Sensors GmbH
Current assignee: Hensoldt Sensors GmbH
Priority date: 2016-10-05
Filing date: 2017-08-22
Publication date: 2018-04-12
Also published as: IL265191A; EP3523674A1; WO2018065146A1; ZA201901366B; AU2017340675A1; EP3523674B1; AU2017340675B2; JP2020500454A; IL265191B; DE102016011913A1

Abstract

The invention relates to a detector unit for detecting a detection signal from a localized optical source in front of a background (H). The detector unit comprises an image sensor (110) and a signal processing unit (120). The image sensor (110) comprises a plurality of pixels (111, 112,...) in order to detect the detection signal and the background (H). The image sensor (110) is designed to generate an image signal (S) independently for each pixel (111, 112,...) depending on the detection signal and the background (H). The signal processing unit (120) is designed to compare the image signal (S) with a threshold (SW) for each pixel (111, 112,...) and to output event data (125) if the threshold (SW) is exceeded within a duration (?t). The threshold (SW) and the duration (?t) can be adjusted.

Description

Detector Unit and a Method for Detecting an Optical Detection Signal The present invention relates to a detector unit and a method for detecting an optical detection signal, and in particular, to a laser warning device and a muzzle flash detector.
Background Both for the detection of pulsed laser sources (for example, for laser controlled missiles) and also for muzzle flashes very short localized optical flashes are to be detected, namely often in front of a strong background which is irradiated by sunlight. The typical flash duration of light flashes may be in the range of nano-seconds for pulsed lasers, and in the range from about loo ps for detecting a muzzle flash.
A primary limiting factor for detecting such signals during daytime is the shot noise of the existing solar background. Typical reference background light sig-nals are a beach irradiated by the sun or a cloud irradiated by the sun (having albedo values of o.6 to 0.8), or even snowfields which are irradiated by the sun (having an albedo value of up to 1), for example.
Previously known systems for the detection of pulsed lasers and hostile fire are significantly limited regarding the background light, and are often not suited for state-of-the-art applications. However, also acoustic detectors are used for fire detection, but they often generate false alarms (for example, in helicopter appli-cations). The also used radar systems are complex, and in addition have to ac-tively emit signals, which is often not desired.
Thus, there is a need for detectors which are capable to detect very short flashes of light of a very low intensity against a background signal of high intensity, as

- 2 -for example objects irradiated by the sun having albedo values of up to 0.8 (cloud, beach) or more. In addition, there is a need for detectors of laser warning systems or detectors for hostile fire warning systems, which may achieve high positional solution of for example < 1 .
Summary At least a part of the problems mentioned above is solved by a detector unit ac-cording to claim 1, a camera according to claim 15, and a method according to claim 19. The dependent claims relate to the advantageous developments of subjects of the independent claims.
The present invention relates to a detector unit for detecting an optical detection signal from a localized optical source in front of a background, and in particular to a detector unit for detecting short optical detecting signals. Thus, the detect-ing unit may in particular be used for laser warning systems and for the detec-tion of hostile fire, and includes: an image sensor and a signal processing unit.
The image sensor comprises a plurality of pixels in order to detect the detection signal and the background, wherein the image sensor is designed to generate an image signal separately for each pixel depending on the detection signal and the background. The signal processing unit is designed to compare the image signal with a threshold for each pixel, and to output event data, if the threshold is ex-ceeded, within a time duration. Optionally the signal processing unit is integrat-ed with the image sensor.
Here, the threshold comparison may be performed by an analog threshold com-parator per pixel, thus the time duration may result from the time constant of said analog threshold comparator. Hereby it will be achieved that detection is not confined to a fixed time pattern defined by the time duration, but event data may be generated every time, when the threshold will be exceeded within the duration specified by the time constant. The threshold and the time constant are configurable, wherein the threshold may be optionally set individually for each pixel. This way it is possible to compensate the different sensitivities of the indi-

- 3 -vidual pixels caused by manufacturing tolerances. Thus, with an appropriate threshold setting, the same sensitivity may be achieved for all pixels. The event data depend on the detection signal (for example, related to only those pixels which detect optical signals from the localized source).
The (electric) image signal may for example represent an electrical voltage value or a number of charges (for example, generated by a CCD sensor). It is also pos-sible that the image signal indicates a current intensity, as generated by a photo-diode, for example. The background also represents an optical signal which may change during the course of time or according the location. Generally, during the considered time interval of nanoseconds or microseconds, there are very few changes in the natural background light, and are dominated by the shot noise which is determined by the photon statistics of the background signal. In order to detect a source, which is localized in a limited region, the signal processing unit comprises a respective signal processing component for each pixel, which performs an evaluation for each pixel and determines a potential event.
In further exemplary embodiments, the signal processing unit is further de-signed to adjust the time period and/or the threshold in a way that a number (rate) of event data output in a predetermined time unit (e. g., per second) is in a predetermined range. Hereby, it may optionally be achieved that the average number of event data per time unit becomes independent of the background.
This way it is achieved that the detector unit is always operated with the maxi-mum sensitivity which is limited by the background radiation.
The present invention is not limited to a specific predetermined range.
Rather, the determined range may be freely chosen and adapted to the respective hard-ware (for example, may only be defined by a maximum value). Thus, for exam-ple, the predetermined range is selected in a way that an optimization of the detection is achieved, and only the number of events (for example, from one source) will be detected, which can be processed. For example, said rate may be about 10 to loo/s for a very simple signal processing hardware; if the best possi-ble sensitivity is desired, a more powerful signal processing hardware may be

- 4 -used, which then also allows to process rates of loo,000/s.
Optionally, the image sensor comprises a CMOS image sensor, and the time period is an integration time for summing up the detection signal.
Optionally, only the signal processing may be realized in a CMOS readout elec-tronics, on which the light-sensitive pixels may be contacted on the backside thereof. This may for example be performed by a so-called "flip chip bonding", where pixel-specific signal components of the readout electronics are contacted on the backside by the image sensor.
In further exemplary embodiments, the signal processing unit is further de-signed to cause a readout of a pixel value for the respective pixel during output of the event data, so that a readout of the pixel is performed in an event-driven and asynchronous way. In particular, no complete frames have to be generated dur-ing readout, instead the pixel values may be processed independently from each another. In order to trigger a readout for said pixel (and only for said pixel), it is sufficient that one single pixel value is exceeded. The pixel value comprises, for example, the summed signal, wherein the summed signal is given by the com-plete number of charges, which have been generated within the time constant of the threshold comparator, for example, and which corresponds to the image value of the respective pixel. Thus, exemplary embodiments also define a signal processing unit, which outputs signal event data separately for each pixel.
In further exemplary embodiments, the detector unit comprises a warning unit, which is designed to generate a warning signal as warning of a detection signal based on the event data. Thus, event data are not necessarily a warning signal.
Only after other criteria are met, a warning signal may be generated therefrom, for example.
As the event data also includes the identification of the pixel, for which the event has been triggered, the warning unit may also determine the direction of the optical source. Hereby, a respective calibration of the image sensor may be used

- 5 -in order to assign a direction in the target region to each pixel, and thus to de-termine the direction of the localized source.
The warning unit optionally comprises a filter which is designed to filter a peri-odic signal by using a first threshold and/or a non-periodic signal by using a second threshold, wherein the first threshold is smaller than the second thresh-old. Hereto, the filter may comprise a digital filter, for example, and the warning unit may comprise a feedback to the signal processing unit in order to cause a change of the threshold.
Optionally, the signal processing unit and/or the warning unit is designed to adjust the threshold in a way that the signal-to-noise-ratio is at a value of (also a smaller or larger ratio may be chosen).
The invention also relates to a camera including a detector unit which has been described above. Optionally, the camera comprises a lens, a spectral filter and/or an interference filter.
The invention also relates to a method for detecting a detection signal of a local-ized optical source in front of a background, wherein the detection signal and the background are detected by an image sensor including a plurality of pixels.
The method comprises the steps: Generating of an image signal, wherein the image signal is generated independently for each pixel, comparing the image signal with a threshold; and outputting event data, if the threshold is exceeded within a time period. Comparing is performed separately for each pixel in an independent way, and the threshold and the time period may be set. Optionally, an individual threshold and/or an individual time period may be set for each pixel.
Exemplary embodiments solve at least some of the technical issues mentioned above by using a CMOS image sensor, in which an analog comparator is inte-grated in each pixel, which compares the detected image signal with the thresh-old and triggers an event, if the threshold is exceeded. Said event indicates for example, that an absolute speed of change of the number of photo electrons has =

- 6 -been exceeded. The content of the respective pixel may then be read out in an asynchronous, event-driven way - not necessarily frame-based.
Here, the threshold may be controlled and set in a way that constant (having a tolerance range of +/- io%, for example) and easily to process event rate is oh-tamed. Said event rate may lie in a range of 10 to 100,000 events/second, for example. However the invention is not to be limited to these values. The actual value to which the event rate is adjusted, may be freely chosen, and may be adapted to the respective situation (for example, computing performance).
Optionally, also event data for each pixel may be counted in signal processing 1 o during fixed time intervals. The thus received obtained number of events per pixel will then be readout at the end of the time intervals for all pixels.
According to further exemplary embodiments, the detector may be formed as a camera, wherein a respective bright lens including an adequate field of vision of 900*900, for example, and a suitable spectral filtering is used.
Short Description of the Figures The exemplary embodiments of the present invention will be better understood by referring to the following detailed specification and the appended figures of the various exemplary embodiments, however it is not intended to limit the dis-closure to the specific embodiments, but they shall only provide a description and facilitate understanding.
Fig. 1 shows a detector according to an exemplary embodiment of the pre-sent invention.
Fig. 2 shows an example of a signal waveform for a signal which is detected by one of the pixels of the image sensor.
Fig. 3 shows the detector including further optional components according to further exemplary embodiments.

,

- 7 -Fig. 4 shows a flow diagram for a method for detecting a detection signal from a localized optical source according to an exemplary embodi-ment of the present invention.
Detailed Description The underlying detector concept, as it is realized in exemplary embodiments of the present invention, may be described as follows.
For an opto-electronic detector, the smallest detectable signal is limited by the intrinsic noise of the detector and by the noise of the background signal against which the useful signal is to be detected. For very short useful signals, the back-ground signal is generally hardly changing during the duration of the useful signal, so that the shot noise generated by the background intensity becomes decisive for the noise contribution of the background. This is calculated from the photon statistics from the root of the photons received at the pixel during detec-tion time.
For an daylight application, as it is the case during usage of a laser warning sys-tem or for muzzle flash detection, the noise ratio, which is generated by the shot noise, is the significantly prevailing ratio, which thus physically limits the detec-tion limit of the detector.
In order to obtain a very low detection limit, the detector is designed such that the number of background photons detected during the detection period be-comes as small as possible, wherein as many photons of the useful signal as pos-sible are to be detected at the same time.
Here, a first step is to use an imaging detector including as many pixels as possi-ble. When the number of pixels is increasing, the field of vision of a single pixel becomes smaller and smaller, thus the light intensity of the background light is reduced. However, as the useful signal to be detected comes from a point-type light source, it remains approximately the same for the hit pixel, irrespective of

- 8 -the number of pixels - a high number of pixels thus reduces the background signal using the same optics without impacting the useful signal.
A second step is the reduction of the detection period. The shorter the detection period is, the less photons of the background light are detected. As long as the detection period is larger than the duration of the useful signal, it will be detect-ed in the complete range thereof. Ideally, the detection period should be the same as the duration of the useful signal. In conventional CMOS cameras, the image frequency is about 50 to 200 Hz, the detection period is between 5 and ms and thus by a factor of o.5-2*1o6 longer than a duration of a laser pulse of for example 10 ns. Even very sophisticated high-speed cameras achieve image fre-quencies of only up to about loo kHz, whereby the detection period corresponds to 10 ms and is thus by a factor of 1,000 longer than the duration of the above-mentioned laser pulse. Thus the known imaging sensors, where the complete image is read out of after integration time, respectively, are not suited very well for the detection of smallest impulses of light against very bright background light. This limitation is circumvented by the proposed signal processing for each pixel. Here, threshold comparators of up to 10 ns time constants may be achieved in CMOS technology. The effective detection time corresponds to this time constant and thus may ideally be realized at the same length as the laser pulse which is to be detected.
The third step is a spectral limitation of the detected wavelength range by means of a respective interference filter. As long as a useful signal lies within the detec-tion band, it is not impacted thereof, however the background signal decreases corresponding to the band width. However, this spectral limitation may only be used insofar as the wavelength of the useful signal is known. In addition, a min-imum bandwidth is limited by the desired viewing angle of the camera, as inter-ference filters comprise a wavelength shift with the incidence angle. This limita-tion may be reduced by using dedicated lenses, as for example the one shown in Fig. 3.
In order to limit the triggering number of false alarms, a relatively high detection

- 9 -threshold of for example S/N = 13 may be used (S/N = signal to noise ratio).
If the useful signal is periodic, also significantly smaller detection thresholds may be used. These first generate a relatively high background signal rate, from which then the existence of the periodic signal may be detected by using a digital comb filter, for example, by thus triggering only a few false alarms.
In the following an actual detector and a camera which realize said detector concept will be described.
Fig. 1 shows an exemplary embodiment of a detector which is capable of detect-ing a detection signal of a localized optical source (not shown in Fig. 1) in front of a background. The detector comprises an image sensor no, a signal processing unit 120, and a warning unit 130.
The image sensor no comprises a plurality of pixel 111, 112,... for optically de-tecting the detection signal and the background. The image sensor no is de-signed to generate electric image signals for each individual pixel 111, 112,... and is designed as a pixel array of photodiodes based on silicon or InGaAs (or other III-V semiconductor materials), for example, which is coupled to the CMOS
signal processing unit 120 on the back side. The image signals are electrical sig-nals corresponding to the incident light intensity and are readout by the CMOS

signal processing components (121, 122, 123) of the signal processing unit 120.
The signal processing unit 120 is designed to continuously compare the image signal with a threshold for each pixel, and to trigger event data 125, if the thresh-old is exceeded within a time period, which are then output via the output unit 129 to the warning unit 130.
The event data 125 indicate an event which potentially has to be detected, and the output unit 129 may then also output image data (for example, intensity) together with said event data 125 and also other information regarding the re-spective pixel (for example, point of time, when the threshold is exceeded).
The warning unit 130 generates a warning or alarm signal 135, in case predetermined criteria are met. These criteria may for example include that a predetermined

- 10 -minimum number or maximum number of pixels has been detected, when the threshold has been exceeded, and thus a localized optical source has been de-tected, and not only a random noise or a sudden brightening of the complete background has occurred.
Fig. 2 shows by way of example a signal waveform for an image signal S which is detected for one of the pixels by the image sensor no. The image signal S may represent a photoelectric current (from a photodiode) or a CCD pixel in the im-age sensor no, for example. The exemplary signal waveform shows the back-ground signal given by the background light and also two short superimposed useful signals Si and S2. The signal K is the resulting signal at the threshold comparator with the time constant At without the middle background intensity (which may be subtracted, for example). The dotted line represents the thresh-old. At the times ti and t2 the threshold comparator is triggered, respectively, as the integrated signal, which is located in the interval At before ti or t2, is exceed-ing the threshold. By using an analog signal processing, detection is not bound to a time pattern. That means, distances between ti and t2 do not have to be an integer multiple of At. For each time ti and t2 the event signal 125 is triggered for the pixel and transferred to the warning unit. Optionally, also the level of the signal or the time of triggering in relation to a reference counter may be part of the event signal.
If said image signal is only detected by a limited number of pixels, it is to be assumed, that it was generated by a localized source and does not represent a random noise signal and a sudden brightening of the background.
The time duration/integration time At may be selected in a variable way, and may be modified by the signal processing unit 120 or the alerting unit 130 de-pending on predetermined criteria, for example. Accordingly, the signal S is not necessarily summed over a predetermined constant period of time. Rather, the range At may be flexibly adapted. In addition to or instead of the integration time At, the threshold SW may also be adjusted. The threshold may be set by means of a comparator, for example, which may be present in each pixel.

- 11 -In addition, warning unit 130 may comprise an optional filter in order to locate periodic signals of a very weak repeating source. Hereto, a digital filter can be used, for example. For example, such periodic sources are controlled laser for laser guided missiles (Beamrider missiles). An advantage of these filters is that very weak signals can be detected. For example, for this application thresholds SW1 may be used, which are smaller than S/N = 13. For non-periodic sources, the warning unit is capable of looking for signal intensities of pixels and would thus apply a higher threshold SW2 (for example, a value of S/N = 13 or more).
This allows to generate an alarm for a single event with a low false alarm rate.
1 o Said changes or adjustments of the threshold SW (or SWi for periodic, and SW2 for non-periodic sources) may also be performed dynamically (for example, during operation), and may ensure that predetermined criteria are met. One of the criteria is, for example, the event rate, that is how many events are detected within a predetermined time (for example, per second). Thus, it may be advan-tageous to select the magnitude of a rate in a way (for example, by setting the threshold) that the detected event rate may be processed by the downstream electronics, but not higher. If the threshold SW is continuously (dynamically) adjusted, an approximately constant event rate may be achieved irrespective of the actual background signal. By setting the integration time At, the detector may be optimized for the expected useful signals by setting an integration time corresponding to the expected length of the useful signal.
It is also possible to control the event rate by the threshold SW, so that the event rate does not exceed a maximum value. The maximum value may, for example, depend on the computing power of the downstream evaluation electronics, and may ensure that events will not be ignored or may be evaluated at a point of time which is not acceptable for the specific application (for example, laser warning system or hostile fire detector).
In further exemplary embodiments, the detector unit is integrated in a camera, which for example, has a large aperture, a lens with spectral filtering and a cus-tom CMOS sensor focal plane array. In particular, the applicable wavelength

- 12 -range for this concept covers the UV range, the visible range, and the near infra-red spectrum (NIR), up to a wavelength of for example 1064 nm. In turn, the detector has a threshold detector which is integrated in each pixel. The threshold detector re-triggers, when an configurable threshold signal (e. g., SW) is reached, and namely within its integration time (At). Said event triggers an asynchronous readout of the pixel value, wherein the pixel coordinate, and the pixel intensity may also be provided together with a time value, in order to obtain access to the detected event. The asynchronous readout is thus event-driven and independent of any readouts of other pixels.
In the exemplary embodiment shown in Fig. 1, wherein the pixel array is coupled to a CCD evaluation unit on the back side thereof, also a wavelength of up to about 1.9 microns may be supported by using a pixel array in InGaAs technology.
Fig. 3 shows an exemplary embodiment according to the present invention for said detector/camera containing the further optional components.
The exemplary embodiment shown represents, for example, a camera is, where-in the image sensor no and the processing unit 120 constitute a unit, wherein an optical signal from a localized optical source 50 is projected on a first position Pi on a sensor surface (image sensor pixel no) by means of one or more optical components 140. In addition to the optical signal, the background H will also be projected on the sensor surface of the image sensor no after passing through the optical components 140. By way of example, a part of the background H is pro-jected on a second position P2 of the sensor surface. It is understood that each position Pi, P2, ... on the sensor surface detects another direction of the back-ground, and it is determined by a calibration, which position on the sensor sur-face (e. g., which pixel) "is facing in which direction". Thus, the direction of the detected optical source may be determined by the identification of the pixels.
The image sensor no and the processing unit 120 generate event data 125 based on the detected optical signals, which are forwarded to the warning unit 130.
The warning unit 130 includes a warning processor to generate a warning signal

- 13 -or an alarm 135 based on the event data and other criteria, for example. For example, the event data 125 comprise the image values, which have been detect-ed by the individual pixels and which correspond to the respective intensity of the incident optical radiation. Optionally, also the detected time and the position of the pixel may be transferred as part of the event data 125 in order to identify the direction of the optical source based thereon, as described above.
In order to perform a narrow spectral filtering and thus achieve a desired sensi-tivity, the detector may include a special lens design. For example, the optical component(s) 140 comprise(s) a lens system of an inverted Galilean telescope 141 for an angular spread, an interference filter 142 and a primary lens 143.
The interference filter 142 is thus inserted at a position having a low beam diver-gence, so that the wavelength shift of the transmission is reduced by the angle of incidence. A suitable lens design would allow the use of a 40 nm filter width for a 900 nm laser detection in a 90 -FOV camera, for example.
In particular, embodiments enable a laser warning system with only minor re-curring costs. In addition, different individual line filter sensors may be used for the laser warner in order to improve the sensitivity and to provide wavelength information in addition to the warning. Detection sensitivities are achieved, which could not be realized with previously known laser detectors having a high positional resolution.
For any values specified in this application, it is to be understood that any varia-tions within a tolerance range may also fall within the defined objects and are considered as disclosed. Thus, an upper tolerance limit may deviate by + 10%
or + 50% or + 100 %, and the lower tolerance limit may deviate by -10% or 30% or -50%) of the specified value.
The features of the invention disclosed in the specification, the claims and the figures may each be individually or in any combination substantial for realizing the invention.

14 -List of reference numbers 50 Optical source no Image sensor 111, 112, 113, ... .. Pixel 120 Signal processing unit 125 Event data 130 Warning unit 135 Warning signal 140 Optical parts 141 Galilei filter 142 Interference filter 143 Primary lenses Background SW, SWi, SW2 Threshold P1, P2 Positions on the sensor surface

Claims

1. Detector unit for detecting a detection signal from a localized optical source (50) in front of a background (H) comprising the features:
an image sensor (110) including a plurality of pixels (111, 112,...) for de-tecting the detection signal and the background (H), wherein the image sensor (110) is designed to generate an image signal (S) separately for each pixel (111, 112,...) depending on the detection signal and the back-ground (H); and a signal processing unit (120), which is designed to compare the image signal (S) with a threshold (B) for each pixel (111, 112,...), and to output event data (125), if the threshold (SW) is exceeded within a duration (.DELTA.t), wherein the threshold (SW) and/or the duration (.DELTA.t) is configurable de-pending on the pixels.

2. Detector unit according to claim 1, wherein the signal processing unit (120) is integrated with the image sensor (110) and the threshold (SW) is configurable separately for each pixel (111, 112,...), and the detector unit further comprises an analog comparator for each pixel (111, 112, ...) in or-der to compare the image signal (S) with an individual threshold (SW) for each pixel (111, 112, but...) by using the analog comparator, wherein the duration (.DELTA.t) is an integration time for summing up the image signal (S) result from the time constant of the analog comparator.

3. Detector unit according to claim 1 or claim 2, wherein the signal pro-cessing unit (120) is further designed to adjust the duration (.DELTA.t) and/or the threshold (SW) in a way that an average number of event data (125), which are output per time unit, lies within a predetermined range.

4. Detector unit according to claim 3, wherein the signal processing unit (120) is further designed to continuously adjust the duration (.DELTA.t) and/or the threshold (SW) in a way that the average number of event data per time unit is independent of the background (H).

5. Detector unit according to one of the preceding claims, wherein the signal processing unit (120) is further designed to derive the thresholds (SW) for each pixel (111, 112, ...) from a global threshold and thus enable each pixel (111, 112, ...) to have the same light sensitivity and to compensate for vari-ances of sensitivity of the pixels (111, 112,...) due to manufacturing rea-sons.

6. Detector unit according to one of the preceding claims, wherein the image sensor (110) comprises a CMOS image sensor.

7. Detector unit according to one of the preceding claims, wherein the signal processing unit (120) comprises a CMOS readout electronics, which con-tacts the image sensor (110) on the backside pixel by pixel by means of pixel-specific signal components (121, 122, 123, ...).

8. Detector unit according to claim 7, wherein the pixels (111, 112,...) of the image sensor (110) comprise photodiodes based on silicon or InGaAs.

9. Detector unit according to one of the preceding claims, wherein the signal processing unit (120) is further designed to cause a readout of a pixel val-ue for the respective pixel (111, 112,...) when outputting the event data (125), wherein the readout is performed in an event-driven and asynchro-nous way.

10. Detector unit according to one of the preceding claims, wherein the signal processing unit (120) is designed to output single event data (125) sepa-rately for each pixel (111, 112, ...) independently of each other.

11. Detector unit according to one of the preceding claims, wherein the signal processing unit (120) is further designed to count the event data which have been obtained for each pixel (111, 112,...) in irregular intervals, and to output the obtained results for all pixels (in, 112,...) at the end of each time interval, respectively.

12. Detector unit according to one of the preceding claims, which further comprises a warning unit (130), which is designed to generate a warning signal (135) as a warning of the localized optical source (50) based on the event data (125).

13. Detector unit according to claim 12, wherein the warning unit (130) com-prises a filter, and the filter is designed to filter a periodic signal by using a first threshold (SW1) and/or a non-periodic signal by using a second threshold (SW2), wherein the second threshold (SW2) is larger than the first threshold (SW1).

14. Detector unit according to one of the preceding claims, wherein the signal processing unit (120) and/or the warning unit (130) are designed to set the threshold (SW) in a way that a signal-to-noise-ratio is larger than a value of 13.

15. Camera comprising a detector unit according to one of the claims 1 to 14.

16. Camera according to claim 15, wherein the field of vision of one pixel (111, 112, ...) is at maximum to 2°.

17. Camera according to claim 15 or claim 16, which further comprises a lens, a spectral filter and/or an interference filter and in particular detects a field of vision of at least 60°*60°.

18. Warning device, in particular a laser warning system or a muzzle flash detector including a detector unit according to one of the claims 1 to 14 and/or a camera according to one of the claims 15 to 17.

19. Method for detecting a detection signal of a localized optical source (50) in front of a background (H), wherein the detection signal and the back-ground (H) are detected by an image sensor (11o) including a plurality of pixels (111, 112, ...), comprising the steps of:
generating (S110) of an image signal (S), wherein the image signal (S) is generated independently for each pixel (111, 112,...);
comparing (S120) of the image signal (S) with a threshold (SW); and outputting (S130) of event data (125), in case the threshold (SW) is ex-ceeded within a duration (.DELTA.t), wherein comparing (S120) is performed separately for each pixel (111, 112,...), and the threshold (SW) and the duration (.DELTA.t) are configurable.

20. Method according to claim 19, which further comprises changing of the threshold (SW) and/or the duration (.DELTA.t) in order to change an average rate of the output event data (125) and, in particular, to keep the average rate below a maximum value.