Persistent filtering in SPD arrays

Abstract

The present invention relates to a system, detector element and method for high speed imaging with an SPD array, as well as a calibration routine for the SPD array.

Description

[Technical field]

[0001] The present invention relates to an improved method and system for high speed imaging based on single photon detectors (SPDs) in which false positives are removed from the results. [Background technology]

[0002] In prior art active imaging scanning, a light beam, typically a laser, is moved over the area to be captured and the position where the beam strikes at each moment is recorded via several image sensors. By processing the difference in position from different viewpoints (sensors), the effective distance to the object illuminated by the beam can be determined by triangulation. Such measurements result in the capture of voxels. The speed at which this processing can be performed, the voxel rate, is on the one hand limited by the speed at which the scan with the light beam is performed, but on the other hand (most strongly) limited by the processing time required by the sensor for the detection of the reflected light beam, especially in relation to background radiation (extraneous light) and thermal noise in general. By specifically addressing this second problem, imaging can be significantly sped up.

[0003] Thus, prior art high speed imaging systems have to address many requirements and challenges. The first challenge is the detection accuracy of the incident photons. Higher and higher spatial resolution is required. As a result, the sensor array is subdivided into more and more individual detectors (pixels). This means that the computational power required to read all these detectors and process the output signals into an image increases. Since the individual pixels (must) always get smaller, they must have high light sensitivity, which limits the spatial window for capturing photons. In addition, a certain processing speed must be achieved. To achieve voxel rates of tens to hundreds of millions of voxels per second, each voxel needs to be captured in 10 ns or less. Therefore, the sensor must also be suitable to operate with a limited photon budget (i.e. the number of detected photons incident on the sensor that are sufficient for detection), e.g. 10 photons. With a limited processing time, only a limited number of photons can be collected.

[0004] For this reason, highly sensitive detectors such as SPDs and SPADs (single photon avalanche detectors) are often considered.

[0005] However, such detectors have the inherent drawback that their high light sensitivity results in unwanted excitation due to both ambient light and thermal noise, which, as such, at the detector level is difficult to distinguish from "real" detections resulting from the incidence of reflected light intentionally emitted by the system to scan the scene.

[0006] Currently, no sensor architecture system or method exists that can reach the desired detection rate under proposed characteristics such as the photon budget, while at the same time guaranteeing a minimum number of false positive detections.

[0007] The object of the present invention is to find a solution to at least some of the aforementioned problems. Summary of the Invention

[0008] The present invention relates to an improved high-speed imaging method as claimed in claim 1.

[0009] Where conventional SPDs, and in particular SPAD arrays, have to deal with a very high number of false positives, especially due to thermal noise and ambient light, the aim of the invention is to remove said random false positive signals. In conventional systems, each photon incident on an SPD or SPAD generates a detection signal. The components of the system are specially designed to record incidences even at low photon densities, which gives advantages in speed and resolution. The smaller the surface of the SPD / SPADs, the more SPDs / SPADs per surface there are, the higher the resolution, but the lower the probability of photon incidence, so the light sensitivity of the components must be increased and they must be triggered faster. As a result, false positives are inevitable in most systems. The aim of the invention is therefore to recognize and remove these false positives from the results, and to see only "real" photon incidences (due to active illumination of the object or scene by the system's light source).

[0010] Preferred embodiments are set out in the following claims.

[0011] In a second aspect, the present invention relates to a detector element and sensor system for high speed imaging, preferably suitable for carrying out the method according to the first aspect, as defined in claims 12, 13 and 14. [Brief description of the drawings]

[0012] [Figure 1A] FIG. 1A shows the time course of an irradiation pattern. [Figure 1B] FIG. 1B illustrates the detection of photon incidence with an active illumination signal. [Figure 1C] FIG. 1C illustrates detection of photon incidence without an active illumination signal. [Figure 2A] FIG. 2A shows the irradiation pattern over time. [Figure 2B]FIG. 2B illustrates the detection of photon incidence with an active illumination signal. [Figure 2C] FIG. 2C illustrates detection of photon incidence without an active illumination signal. [Diagram 3] FIG. 3 is a diagram illustrating a circuit for imposing a first persistence condition according to an embodiment of the present invention. [Figure 4] FIG. 4 illustrates a circuit for imposing a second persistence condition according to an embodiment of the present invention. [Diagram 5] FIG. 5 illustrates a circuit for imposing a third persistence condition according to an embodiment of the present invention. [Figure 6] FIG. 6 is a diagram illustrating a circuit for imposing a third duration condition in accordance with an embodiment of the present invention, combined with a calibration circuit. [Figure 7] FIG. 7 illustrates a circuit for imposing a third persistence condition in accordance with an embodiment of the present invention, combined with a calibration circuit and a local match circuit. [Figure 8A] FIG. 8A is a diagram illustrating a first calibration circuit according to an embodiment of the present invention. [Figure 8B] FIG. 8B shows a circuit for imposing a persistence condition according to an embodiment of the invention, combined with a calibration circuit according to FIG. 8A. [Figure 9A] FIG. 9A illustrates a second calibration circuit according to an embodiment of the present invention. [Figure 9B] FIG. 9B shows a circuit for imposing a persistence condition according to an embodiment of the invention, combined with the calibration circuit according to FIG. 9A and the calibration circuit of FIG. 8A. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] Unless otherwise defined, all terms used in the description of the present invention, including technical and scientific terms, have the meanings commonly understood by those skilled in the art of the present invention. In order to better appreciate the description of the present invention, the following terms are explicitly explained.

[0014] "A," "an," and "the" are used herein to refer to both the singular and the plural, unless the context clearly indicates otherwise. For example, "a segment" means one or more than one segment.

[0015] When "approximately" or "around" is used herein with respect to a measurable quantity, parameter, duration or moment, etc., it means a variation relative to the cited value and within ±20%, preferably within ±10%, more preferably within ±5%, even more preferably within ±1%, even more preferably within ±0.1% of the cited value, insofar as such variations are applicable in the described invention. However, it should be understood that the value of the quantity where the term "approximately" or "around" is used is itself specifically disclosed.

[0016] The terms "comprise," "comprising," "consist of," "consisting of," "provided with," "contain," "containing," "include," "including," "hold," and "holding" are synonymous and are inclusive or open-ended terms indicating the presence of what follows without excluding or excluding the presence of other components, features, elements, materials, or steps that are known from or described in the prior art.

[0017] The term "pixel" herein refers to a light-sensing unit or photodetector, preferably in the form of an SPD or SPAD. Typically, said pixels are provided in a matrix-like array.

[0018] The term "macropixel" refers to a virtual pixel composed of several pixels or detectors, typically forming a contiguous cluster. The macropixel is associated with a specific set of spatiotemporal conditions and may vary in size, shape and other parameters depending on the situation. Macropixels may overlap each other and have common pixels or detectors. Preferably, macropixels have a square or rectangular shape, but may also have a generally circular or elliptical, triangular, cross or star shape or other shapes. A macropixel is composed of at least two detectors (e.g. in a 2x1 or 1x2 matrix shape), more preferably at least four detectors (e.g. 2x2).

[0019] Recitation of a numerical interval by endpoints includes those endpoints, and all integer, fractional, and / or real numbers between the endpoints.

[0020] In a first aspect, the present invention relates to a method for high speed imaging based on an array having a plurality of Single Photon Detectors (SPDs), preferably Single Photon Avalanche Detectors (SPADs), comprising: a. capturing a photon on an SPD, whereby said SPD generates and supplies a detection signal C_ov to an evaluation circuit, whereby said detection signal describes a positive observation state when observation by said SPD of the electromagnetic radiation to be detected occurs and describes a negative observation state when observation by said SPD of the electromagnetic radiation to be detected does not occur; b. evaluating the detection signals of the SPDs by an evaluation circuit for each SPD, and determining that the detection signal for each SPD must satisfy a persistence condition in order to generate for the SPD a confirmation signal for the electromagnetic radiation to be detected; has.

[0021] The persistence condition for at least one, and preferably all, SPDs requires a positive observation state of the SPD in at least M, preferably consecutive, observation windows, preferably of period T, out of the past N observation windows, where M is greater than 1 and at most equal to N.

[0022] By looking at the observation states in the past N observation windows, it can be verified whether the positive observation state is persistent and confirmed by previous measurements. In this way, false positives due to thermal noise etc. are filtered out of the result. The detection is confirmed based on a confirmation signal only if a certain number or a certain percentage of the past N observation windows also show a positive observation state. The evaluation can be performed locally for each pixel or SPD / SPAD or centrally for all (or a part of) the SPD / SPAD in an evaluation component or evaluation circuit.

[0023] It should be understood that the duration condition of M (consecutive or non-consecutive) observation windows out of N can be applied in alternative ways, in particular using a ratio M / N of the past N observation windows (again with or without the condition that the M are consecutive, where M is typically equal to N).

[0024] Of course, more specific persistence conditions can be applied as desired (e.g., at least M of the past N observation windows are positive observation states, of which at least P are consecutive positive observation states; there are up to Q consecutive non-positive observation states; there are at least 2, 3, ... clusters of P consecutive positive observation states, etc.). Combinations and variations thereof are also self-evidently within the scope of the present invention.

[0025] The above method allows for consistent detection of "active" radiation because such radiation typically impinges broadly over a given area for a given period of time and therefore satisfies the persistence condition, unlike thermal noise and other unwanted detections.

[0026] In a first preferred embodiment, the method comprises illuminating a scene (such as an object or an environment to be imaged) with one or more scanning light sources (lasers, LEDs, etc.) according to a predefined illumination pattern, said illumination pattern comprising light pulses separated in time (with or without regular length and / or spacing).

[0027] The observation window on which the duration condition is imposed is time-related to one of the light pulses, and the detection signal indicates the presence or absence of an observation by the SPD (positive observation state or not).

[0028] Temporally relating the observation windows to the light pulses means timing the observation windows so that they approximately overlap (in start and / or end times), which allows the radiation observed in the observation windows to be linked with a very high probability to the light pulses (and therefore not to thermal noise or other noise signals).

[0029] The applicant has realised that a balance needs to be found between the number of light pulses in the illumination pattern (and the length of the pulses and the time between pulses) and the desired speed at which the system can operate. A larger number of pulses / longer pulse lengths / longer spacing means that it will take longer to evaluate the signal on the SPD from all the discrete positions. For this reason, N is typically set between 2 and 10, preferably between 2 and 7, more preferably between 2 and 5, e.g. 2, 3, 4 or 5.

[0030] In a preferred embodiment, the observation window is set to substantially overlap with the light pulse with which it is associated (i.e., substantially covering the duration of the light pulse, but not necessarily so limited) and have a substantially coincident start time (whereby the start time of the observation window is at the earliest simultaneous with the start time of the light pulse), whereby substantially coincident means that the start time of the observation window is within 0.0 ns to a maximum of 1.0 ns, preferably within a maximum of 0.5 ns, more preferably within a maximum of 0.1 ns, or even within 0.05 ns after the start time of the light pulse.

[0031] As a result, it is ensured that (substantially) all radiation impinging during the observation window is due to an event during the light pulse.The start time allows for play to take into account time-of-flight.

[0032] In a preferred embodiment, the detector comprises a quenching circuit, more preferably a passive quenching circuit, or alternatively an active quenching circuit.

[0033] In a preferred embodiment, the observation window is set to an end time that is between 0.0 ns and 2.0 ns after the end time of the associated light pulse, preferably between 0.05 ns and 15.0 ns, more preferably between 0.1 ns and 10.0 ns after the end time of the associated light pulse, such timing ensures that radiation due to events during the light pulse is captured by the observation window taking into account time of flight, as explained above.

[0034] By simultaneously applying the above-mentioned restrictions on the start and end times, the number of photons that are not attributable to the light pulse and are captured during the observation window can be reduced as much as possible.

[0035] In a second alternative preferred embodiment, the method comprises illuminating the scene with one or more scanning light sources, whereby said scanning light sources illuminate said scene substantially continuously, meaning that said scanning light sources illuminate continuously and with a substantially constant intensity, in which case the observation window can be selected more arbitrarily, possibly even continuously.

[0036] The light source is used to emit an active and controlled light signal, thereby illuminating a scene / object on which an image is to be formed, and the SPD forms an image of the illuminated scene / object by capturing reflected photons and recording the incidence on a further component based on a detection signal.

[0037] Thus, the method includes illuminating a scene or object with the scanning light source(s), which illuminates in discrete or quasi-discrete steps, and for each illuminated discrete or quasi-discrete location, continues illuminating for a short time before moving on to the next discrete location. Indeed, in this specification, the term "discrete location" refers to a discrete direction of the scanning light source, i.e., a discrete location that is illuminated. The "path" of the discrete locations followed during illumination can be an irregular path, but typically follows a row-by-row or column-by-column scanning pattern. However, it should be noted that the invention is in no way limited to this aspect. Thus, "quasi-discrete" refers to a situation where the light source keeps changing its direction, but the change is made at such a speed that for (two or more) consecutive observation windows, the covered distance of the illuminated area is so limited that it can be considered as the same discrete area. Thus, in the following description, "discrete" should be interpreted to include quasi-discrete.

[0038] Preferably, M and / or N are adjustable, more preferably for each SPD. In certain embodiments, M and / or N are dynamically configurable, and the system can automatically adapt to a large or small number of noise detections (detected signals that do not satisfy the persistence condition) for an SPD. This allows for faster and more accurate operation by relaxing the persistence condition for SPDs with statistically little observed noise and tightening the persistence condition for SPDs with statistically observed high noise.

[0039] Thus, the ability to individually tune M and / or N can be used to make a particular SPD more or less sensitive, for example when a large number of detection windows are detected with no active light detection signal, which may indicate a "leaky pixel", or can be turned off completely if desired.

[0040] In a preferred embodiment, M and / or N are adapted automatically based on a calibration routine.

[0041] In a preferred embodiment, the persistence condition is combined with the fulfillment of a coincidence condition, which is based on the assumption that, in the presence of an observation by an SPD, the incidence of a photon on a particular SPD is not a single shot, but involves the incidence of a spatially clustered photon flux. Based on this, a coincidence condition is imposed that verifies the observation state or confirmation signal of a neighboring SPD within a certain time around a positive observation state or confirmation signal of a certain SPD.

[0042] The match condition can be imposed before or after the persistence condition, which is more efficient depending on the circuit.

[0043] In this way, the first SPD that satisfies the persistence condition can subsequently be subjected to the match condition, and a confirmation signal is generated only after the match condition is satisfied. Preferably, the match condition implies that at least one neighbouring SPD (adjacent to the first SPD) also satisfies the persistence condition. Obviously, this allows narrowing the number of neighbouring SPDs that need to satisfy the persistence condition to at least 2, 3, 4 etc.

[0044] The term "adjacent SPDs" primarily refers to directly adjacent SPDs, which are typically arranged in a rectangular array of rows and columns, meaning that adjacent SPDs are located in adjacent columns of one and the same row, or adjacent rows of one and the same column. However, the term can simply be expanded to include, for example, diagonally adjacent SPDs (adjacent rows and adjacent columns), or can be further expanded such that "diagonally adjacent" is also included herein.

[0045] The matching condition can be imposed in many ways and with variable conditions, so that the number of adjacent SPDs that satisfy the persistence condition can be adapted (1, 2, 3, 4, 5, 6, 7, 8, 9 or more), but also the definition of "adjacent". Thus, in a preferred embodiment, adjacent can be interpreted as "directly adjacent", but also "diagonally adjacent" SPDs can be accepted (directly adjacent to the first directly adjacent SPD in another dimension than the first SPD is directly adjacent), which can also be defined as being located within a predetermined interval of the first SPD (thus also allowing "opposite proximity"). This can also be changed depending on the situation, and the system can dynamically adapt it (e.g. via a calibration routine).

[0046] An initial filtering is performed by first imposing the persistence condition, and then the coincidence condition is applied only to the remaining detections that were confirmed according to the persistence condition. This sequence ensures that the computationally most difficult condition (i.e. the coincidence condition, since it has to consider the signals of a large number of adjacent SPDs) is applied to a smaller number of signals.

[0047] In the first method, the SPDs adjacent to a first SPD with location (i,j) are simply defined as having coordinates (i,j±1) or (i±1,j). To add diagonally adjacent SPDs, SPDs with coordinates (i±1,j±1) are also accepted.

[0048] If necessary, the range of "adjacent SPDs" can be extended to SPDs with coordinates (i±2,j±1), (i±1,j±2), (i±2,j), (i,j±2), and even to (i±2,j±2).

[0049] Of course, it is suggested that further extensions are possible and are considered part of the present invention.

[0050] Therefore, the match condition based on the definition of neighboring SPDs can also be adapted, whereby all SPDs are weighted based on their "distance" in the array to the first SPD, and only when a certain weight is reached for an SPD for which the persistence condition is satisfied, the match condition is considered satisfied for the first SPD.

[0051] Conversely, the consistency condition can be imposed on the SPDs first (in the manner described above, but instead of satisfying the persistence condition, it is imposed for each observation window based on the positive observation states in the observation window), and only those SPDs for which the consistency condition is satisfied are subsequently checked for satisfaction of the persistence condition.

[0052] In a second aspect, the invention relates to a detection element comprising a Single Photon Detector (SPD), preferably a Single Photon Avalanche Detector (SPAD), configured to generate a signal C_ov describing the state of observation or non-observation of electromagnetic radiation (photons) to be detected during successive observation windows, a memory element configured to store the state of the SPD at the end of at least N previous observation windows (N being a natural number greater than or equal to 1), and a logic circuit configured to generate a confirmation signal if the signal of the SPD at the end of a current observation window and the signals of the SPD of at least M observation windows (M being a natural number greater than or equal to 1 and at most equal to N) of the N previous observation windows indicate observation by the SPD.

[0053] In such a case, the detector elements may individually impose persistence conditions on their own observations and generate and send a confirmation signal to the central processor only if the persistence conditions are met. Examples of this possible embodiment are further described in the text and figures.

[0054] Alternatively, to further simplify the detection element, the persistence condition is implemented centrally.

[0055] In a preferred embodiment, the spacing between successive light pulses is at least 0.5 ns, preferably at least 1.0 ns, more preferably at least 2.5 ns, and most preferably at least 5.0 ns.

[0056] In a preferred embodiment, successive light pulses are spaced apart by less than 1000 ns, preferably less than 500 ns, even more preferably less than 100 ns, and most preferably less than 50 ns.

[0057] Most preferred is a range that combines the above minimum and maximum values.

[0058] In a preferred embodiment, the length of the light pulses of a given illumination pattern and the length of its associated observation window have a ratio between 1.0 and 2.0, preferably between 1.0 and 1.5, and most preferably between 1.0 and 1.25, to ensure that as much of the signal resulting from the light pulses is captured in the observation window as possible without making it unnecessarily long (to the point that the method becomes slow).

[0059] In a preferred embodiment, the observation states of the past N observation windows are stored by memory elements on the SPADs or on a central processing unit (which imposes the persistence condition on some or all of the SPADs in the array), and the "old" observation states (from observation windows older than N) are overwritten by new observation states. Such memory elements may include flip-flops, DRAMs (dynamic RAM), SRAMs (static RAM), analog memory elements, or other components.

[0060] By linking the observed state to an appropriate logic circuit (e.g. a modified AND port), one ensures in a simple way that the desired processing of the signal is achieved and that confirmation is obtained only if the persistence condition is met.

[0061] A buffer element is preferably provided in front of the logic circuit (and memory element) to separate the operation of the SPAD from the signal processing.

[0062] In a particular embodiment, for example when N is simply equal to 2, a so-called delay circuit can be provided which includes the signal from the previous observation window and feeds it together with the signal of the current observation window to a logic circuit such as a simple AND port. Since the observation windows typically occur at regular time intervals, such a circuit can be realized in a simple manner without requiring many additional components. Also, the persistence condition can be applied locally to the SPAD itself, so that only a (non-)confirmation signal is fed to further processing equipment.

[0063] Many possible circuits for the above purposes are shown in the figures, possibly in combination with further adjustments and refinements.

[0064] In a third aspect, the invention relates to a sensor system for high speed imaging, comprising a matrix / array of detector elements according to the second aspect of the invention and a circuit for reading said detector elements in said matrix / array. By providing said detector elements with their own evaluation circuit, part of the computational load can be decentralized in a simple manner, with only confirmation signals being transmitted to a central read-out circuit.

[0065] Alternatively, in a third aspect, the present invention provides a method for producing a composition comprising the steps of: a. a matrix / array of detector elements having single photon detectors (SPDs), preferably single photon avalanche detectors (SPADs), configured to generate a signal C_ov describing the observed or not observed state of electromagnetic radiation (photons) to be detected during successive observation windows; b. a logic circuit reading the signal C_ov of the detector elements in said matrix / array and having one or more memory elements; The present invention relates to a high-speed imaging sensor system having a

[0066] wherein the one or more memory elements are configured to store a state of the SPD at the end of at least N previous observation windows, N being a natural number greater than or equal to 1. The logic circuit is configured to generate a confirmation signal to the SPD if the signal of the SPD at the end of a current observation window and the signals of the SPD for at least M of the N previous observation windows indicate observation by the SPD, M being a natural number greater than or equal to 1 and at most equal to N. In this way, the detection elements are kept simple (and therefore more compact, less expensive, more robust) and the computational load is centrally borne.

[0067] In a preferred embodiment, the sensor system is configured to perform the method according to the first aspect of the invention.

[0068] In a further aspect, the invention relates to a method for calibrating an imaging system having an array with a number of Single Photon Detectors (SPDs), preferably Single Photon Avalanche Detectors (SPADs), comprising monitoring, by an evaluation circuit (either one evaluation circuit or for each SPD, for each group of SPDs or for the whole system), the observation conditions / detection signals over a calibration period in at least two, preferably at least three, calibration windows in said calibration period, preferably separated in time from one another and performed during a period when the array of SPDs is not actively illuminated (i.e. intentionally by a light source controlled by the system).

[0069] A larger number of calibration windows can be set, resulting in a more thorough and reliable classification, e.g. 5, 6, 7, 8, 10 or more calibration windows, etc. The calibration windows can be contiguous or separated from each other.

[0070] Based on the number of calibration windows in which the SPD observations are detected, the SPDs are classified into several categories separately. The number of classifications or the manner of classification can be set according to the situation. Thus, two classes (reliable-unreliable) can be selected, or more than two classes can be selected.

[0071] This classification indicates how sensitive the SPD is to noise etc. and can be used for further processing of the detection signal for imaging. More specifically, the calibration method can be applied in terms of the present invention according to the previous embodiment (before this embodiment) and further processing of the SPD's signal can be adapted based on the classification. More specifically, the persistence condition can be adapted for the SPD based on the aforementioned classification, for example by tightening or relaxing the persistence condition (making M larger or smaller, or deactivating the persistence condition completely or making M=1) and / or deactivating a particular SPD where too large an error is detected (for example a screamer where the positive observation state is statistically indicated too frequently by the detection signal).

[0072] In a first embodiment, the SPDs can be divided into two categories of reliability: high priority (HP) and low priority (LP). HP SPDs are considered more reliable and can operate under less stringent persistence and / or consistency conditions than LP SPDs. In a non-limiting example, LP SPDs have positive observation conditions in at least T% of the calibration window, where T is preferably at least 50%, more preferably at least 66% or 75%.

[0073] Further classifications can optionally be provided (Medium Priority (MP), High Medium Priority (HMP), Low Medium Priority (LMP), etc.), all of which are associated with specific processing methods during imaging (stricter or more relaxed persistence conditions) and are assigned to specific ranges of the proportion of positive observation conditions detected in the calibration window of the SPD.

[0074] The adaptation of the above conditions can vary significantly, such as simply incrementing M by 1 for each decreasing calibration level (going from HP to LP), but it is equally possible to have different adaptations for each changing category.

[0075] Matched persistence and / or coincidence conditions belonging to a particular category are predefined and the evaluation circuitry is configured to execute them appropriately.

[0076] In a preferred embodiment, the observed states of the past N calibration windows are stored by memory elements on the SPADs or on a central processing unit (which imposes the persistence condition on some or all of the SPADs in the array), and the "old" observed states (from calibration windows older than N) are overwritten by new observed states. Such memory elements may include flip-flops, DRAMs (dynamic RAM), SRAMs (static RAM), analog memory elements, or other components.

[0077] Linking the observed states to appropriate logic circuits (eg, modified AND ports) ensures in a simple manner that the SPD is classified as desired.

[0078] In a particular embodiment, for example when N is simply equal to 2, a so-called delay circuit can be provided that includes the signal from the previous calibration window and feeds it together with the signal of the current calibration window to a logic circuit such as a simple AND port. Typically, the observation windows occur at regular time intervals, so such a circuit can be realized in a simple manner without requiring many additional components. Also, the calibration can be applied locally to the SPAD itself.

[0079] Many possible circuits for the above purposes are shown in the figures, possibly in combination with further adjustments and refinements.

[0080] A wide variety of implementations are possible to hold the state information of all pixels from the past observation window, for example using flip-flops, but are not limited to the use of flip-flops, but can also use memory elements such as SRAM and / or DRAM elements, and in some cases simple parasitic capacitors that can store load.

[0081] The present invention is illustrated below by non-limiting examples which illustrate the invention but are not intended, and should not be construed, as limiting the scope of the invention. EXAMPLES

[0082] 1A to 1C show emitted and received signals according to a possible embodiment of the present invention.

[0083] FIG. 1A shows the time course of illumination by a light source on an SPD, where a pulsed illumination pattern is maintained with a square wave (1) having a constant length and period, and during each period the scene is actively illuminated by the light source for a certain period of time and not actively illuminated by the light source.

[0084] Figure 1B shows the events detected for an SPD, where each event (3a, 3b, 3c) represents the incidence of a photon on the SPD. Some of these events are the result of ambient light, random noise or error signals (3b, 3c), while some originate from photons emitted by the light source (3a). In principle, statistically, a first group is randomly distributed, while a second group is during the period when the light source emits light.

[0085] Fig. 1B shows the observation windows (2; W1, W2, W N ) are also shown. The observation windows (2; W1, W2, W N ) has a certain length and period. The observation window (2; W1, W2, W N ) the observation state of the SPD is recorded (detection, which can be the incidence of a photon in the observation window). By illuminating with an active signal, one or more events are likely to be detected in the observation window (corresponding to the illumination square wave).

[0086] Finally, FIG. 1C again shows the observation window and the detected events (3b, 3c), none of which are the result of illumination by the light source.

[0087] As is clear from Fig. 1C, due to the random distribution of the detected events, in some observation windows, the event (3c) still occurs and may be erroneously considered as the detection of an incident photon on the SPD. However, by imposing a duration condition, for example, that a positive observation state must be observed (i.e., an event detected in an observation window) in two consecutive observation windows (or three, or two out of three, etc., depending on the duration condition imposed), it is ensured that the random event mentioned above is no longer evaluated as a "real" detection in the imaging. Fig. 1C indeed shows only one observation window with a positive observation state.

[0088] However, in FIG. 1B, a “real event” (3a) as a result of illumination by an active signal is indeed confirmed, taking into account that a positive observation state is observed in two consecutive observation windows (and also in all observation windows) and the persistence condition is fulfilled.

[0089] Figures 2A to 2C show the same graphs, with the difference that the light is continuous rather than pulsed, as shown in Figure 2 A. Figure 2B shows only the events caused by illumination in the observation window.

[0090] 3 shows a possible circuit of an SPD detection element according to the invention, where an (optional) coincidence condition is also imposed after the persistence condition via a circuit (4) for this purpose: the persistence condition is that two successive observation windows must show a positive observation state.

[0091] The photon detector SPD or SPAD (5) is connected to a voltage V BE A resistor is connected in series between the detector and the photon detector. When a photon is incident on the detector, a breakdown occurs in the detector, and a current flows through the resistor (in other words, a change in the voltage at the point between the detector and the resistor). A circuit with such a resistor is called "passive quenching". It goes without saying that other "quenching" circuits, such as "active quench" circuits, are also applicable.

[0092] This is followed by a pulse detector that converts the measured voltage into a discrete signal (whereby 1 indicates a positive observation state, i.e., the detection of a photon incident during the observation window, and 0 indicates a negative observation state, i.e., no photon was detected).

[0093] Next, observation window click_T n-1 The signal indicating the observation state of is actually delayed through a flip-flop (FF) according to the period of the observation window (by clocking the FF on the rising edge of the observation window), and then, through the AND port, the observation state is nIf both observed states are positive, the AND port delivers a confirmation signal of 1 at its output, which can be used for further signal processing, for example by combining it with a signal from a nearby detector element (6).

[0094] Based on FIG. 3, it can be seen that more observation states can be easily included by providing additional flip-flop components or memory elements, thereby imposing desired persistence conditions via the logic circuitry.

[0095] Figure 4 shows a further adaptation, where the first flip-flop (FF1) is clocked with the detection signal of the SPD (5). If this occurs during the observation window (so that said signal produces a 1), FF1 also clocks a 1. If there is no observation window (signal = 0), a 0 is also clocked as input to the second flip-flop (FF2). The falling edge of the observation window (NOT-observation window) then acts as a clock trigger. This allows the next observation window ("click_T n When the observation state from the previous observation window ("click_T") arrives as data output Q from the first flip-flop, n-1 The observed states of the first flip-flop and the second flip-flop are combined at the AND port of the subsequent stage.

[0096] By adding more flip-flops using delays, further conditions can be imposed, such as three consecutive positive observation states.

[0097] The above-described configuration may also be followed by one or more circuits (4) that impose a matching condition.

[0098] Figure 5 shows a circuit in which the past N observation states are stored in N memory elements (7). The circuit that determines the observation state routes the signal of the SPD (5) to the clock of a flip-flop (FF), so that the signal of the observation window serves as the data input (D) of the flip-flop (FF). This generates a data output (Q) of 1 when the SPD detects the incidence of a photon in the observation window, and 0 when there is no detection.

[0099] The signals from the flip-flops are stored in successive memory elements, which store the signals in a circular manner as a storage device. These memory elements feed the stored signals to a logic circuit which imposes a persistence condition depending on their settings. Those skilled in the art can easily modify said conditions with circuits using simple components, so no further explanation is given here. In FIG. 5, for convenience, an AND port with N inputs is provided, and said persistence condition requires the detection of a photon incidence in N of the past N observation windows for the generation of a confirmation signal (and is fed, for example, to a further processing unit, where a coincidence condition is possibly imposed).

[0100] Figure 6 shows the circuit of Figure 5 with the addition of a calibration circuit (9) operating according to the routine described herein. In this case, based on the results of the calibration routine and the calibration condition, a high priority (HP) or low priority (LP) is assigned to the SPD (5). As a result, in the case of HP, no further persistence conditions are imposed on the observed state and a confirmation signal is generated immediately. In the case of LP classification, stricter persistence conditions are applied, again requiring N positive observed states in the last N observation windows before a confirmation signal is generated. The SPD can also be switched off by the calibration routine via switch (8) if a problem is detected (e.g. saturation of the SPD).

[0101] In Fig. 7, a match condition is imposed locally on the acknowledgement signals before they are forwarded to further processing units. For a detailed development of the match conditions, reference is made in particular to International Patent Application No. PCT / IB2021 / 054688, Belgian Patent Application No. 2020 / 5975 and Belgian Patent Application No. 2021 / 5521. The acknowledgement signals from the persistence circuit of Fig. 6 are processed by a match circuit (10) before being forwarded to the appropriate bus.

[0102] FIG. 8A shows a possible application of a circuit for imposing a first possible calibration condition on the SPD (5) via a calibration circuit (9a), namely the absence of a photon incident over the past N calibration windows. The incoming "click" signal indicates the detection of a possible photon incident (detection signal), which is then sent to the OR port via an inverting circuit along with a signal indicating whether a calibration routine is being executed. If there is a photon incident (1 as detection signal) and the calibration routine is being executed (thus the calibration condition is not met in the calibration window), the combination is simply 0. The output of the OR port is sent to the NAND port along with the inverted Q output of a flip-flop (FF). The output of the NAND port is sent to the data input D of the flip-flop. The flip-flop is clocked based on the start of each calibration window. As long as there is no incoming photon, the signal sent to the data input remains 0, and the inverted Q output (QN) generates a 1, which is looped back to the data input via the NAND port and the "new" signal of the next calibration window, so that the inverted Q output (QN) generates a 1 as long as there is no incoming photon (hence the "click" is 0, which is inverted when a 1 is sent to the OR port, thus sending a 1 to the NAND port).

[0103] As long as the calibration routine continues to run (and ZeroBG_EnableN=1), the flip-flop (FF) will keep iterating through this loop. As soon as a single photon is detected, the flip-flop will produce a 0 as its inverted output, and will keep looping through the inverted NAND port to always produce a 1.

[0104] The resulting signal, ZeroBG, is then used to classify whether the signal from the persistence routine meets the calibration condition or not, and can therefore be used as the HP or LP output via the ports of the circuit of FIG. 8B.

[0105] Figure 9A shows the possibility of providing yet another second calibration circuit (9b) in addition to the calibration circuit of Figure 8A, which further expands the possible classification of SPDs (5). Whereas the calibration circuit (9a) of Figure 8A seeks to determine which SPDs are insensitive (or have low sensitivity) to background radiation (or thermal noise, etc.), the second calibration circuit aims to determine whether a saturated SPD is present (in other words, will always show detection).

[0106] The SPD signal ("click") is inverted and sent to the AND port together with the calibration routine signal. If the calibration routine is running (noSAT_Enable=1) and a photon incidence or photon event is detected ("click"=1), the AND port generates a "0"; if no event is detected, a "1" is generated.

[0107] The signal from the AND port is input to the OR port together with the loop data output Q of a flip-flop (FF). The result of the OR port, together with the clock signal of the calibration window, becomes the data input of the flip-flop.

[0108] If there is no event (signal 1 to the OR port), then automatically a 1 is generated at the data input, and therefore a 1 is output at the data output Q. This 1 data output Q goes back to the OR port (automatically 1 again), and back to the data input, also 1, and so on in this loop. That is, if there is (at least one) calibration window during the second calibration routine where no event occurs, then the calibration circuit will output a 1, which means that the SPD is not saturated.

[0109] If an event occurs in all calibration windows (click=1), the OR port always receives a 0 from the SPD (after the AND port) and then continues to loop unless preceded by a signal 1 to the data output Q of the flip-flop (FF). Therefore, if consecutive events occur, the output of the second calibration circuit produces a 0, indicating that the SPD is saturated. If noSAT is 0 (saturated), an event always occurs, so the signal of ZeroBG is always 0.

[0110] The resulting signal noSAT is then used to classify whether the signal from the sustained condition satisfies the second calibration condition or not, and thus in the circuit of FIG. 9B, together with the first calibration circuit of FIG. 8A and the resulting signal ZeroBG, is used to classify the SPD and adjust the processing of the results.

[0111] Thus, a circuit presented with noSAT equal to 0 (saturated SPD) and therefore ZeroBG equal to 0 will always reject signals from the SPD even if the SPD satisfies the persistence condition.

[0112] On the other hand, in the case of a non-saturated SPD (noSAT=1), ZeroBG determines whether the SPD is treated as a high priority or a low priority.

[0113] Finally, if noSAT is 0 (SPD is saturated), the SPD is also switched off via the additional circuit (8).

[0114] The present invention is not limited to the embodiments described above, and it is envisaged that numerous modifications or variations can be made to the described examples without reassessing the scope of the appended claims.

Claims

1. A high-speed imaging method based on an array having a plurality of light-sensing units (5), comprising: a) capturing a photon on a light-sensing unit (5), whereby said light-sensing unit (5) generates and supplies a detection signal C_ov to an evaluation circuit, whereby said detection signal describes a positive observation state if said light-sensing unit (5) observes the electromagnetic radiation to be detected, and describes a negative observation state if said light-sensing unit (5) does not observe the electromagnetic radiation to be detected; b) evaluating the detection signals of the light-sensing units (5) by an evaluation circuit for each light-sensing unit (5), whereby the detection signals of each light-sensing unit (5) must satisfy a persistence condition in order to generate a confirmation signal for the light-sensing unit (5); In a high-speed imaging method having said persistence condition for at least one, preferably all, light-sensing units (5) requires a positive observation state of said light-sensing unit (5) in at least M, preferably consecutive, observation windows (2) of preferably period T out of the past N observation windows (2), M being greater than 1 and at most equal to N; High-speed imaging method.

2. 10. A method of high speed imaging as claimed in claim 1, comprising illuminating a scene with one or more scanning light sources according to a predetermined illumination pattern having temporally separated light pulses; each of the observation windows (2) is temporally associated with one of the light pulses, and the detection signal indicates the presence or absence of an observation by the light sensing unit (5) as a result of a light pulse associated with the observation window (2); High-speed imaging method.

3. 3. The high-speed imaging method of claim 2, wherein the observation window has a start time that substantially coincides with a start time of the associated light pulse; Substantially coincident means that the start time of the observation window (2) is within 0 ns to a maximum of 1.0 ns, preferably within a maximum of 0.5 ns, and more preferably within a maximum of 0.1 ns after the start time of the light pulse. High-speed imaging method.

4. 3. The high-speed imaging method according to claim 2, wherein the observation window (2) has an end time lying between 0.0 ns and 25.0 ns after the end time of the associated light pulse, preferably between 0.05 ns and 15.0 ns, more preferably between 0.1 ns and 10.0 ns after the end time of the associated light pulse. High-speed imaging method.

5. 10. A method for high speed imaging as claimed in claim 1, comprising illuminating a scene with one or more scanning light sources, the scanning light source substantially continuously illuminates the scene; High-speed imaging method.

6. 2. The high-speed imaging method according to claim 1, wherein M and / or N are adjustable, preferably for each light-sensing unit (5). High-speed imaging method.

7. 2. The high-speed imaging method according to claim 1, wherein when a coincidence condition is met, a confirmation signal is generated for the first light-sensing unit that satisfies the persistence condition; the coincidence condition requires that at least one other light-sensing unit in the observation window meets the persistence condition; the at least one other light-sensing unit is adjacent to the first light-sensing unit that satisfies the persistence condition; High-speed imaging method.

8. 2. The high-speed imaging method according to claim 1, wherein when the coincidence condition and the persistence condition are satisfied, a confirmation signal is generated for the first light-sensing unit; the evaluation circuit is configured to impose the persistence condition after imposing the coincidence condition only on the light-sensing unit (5) that has satisfied the coincidence condition; The coincidence condition includes receiving a detection signal with a positive observation state by at least one other light sensing unit (5) within a certain time interval from receiving a detection signal with a positive observation state by the first light sensing unit (5); the at least one other light-sensing unit (5) is adjacent to the first light-sensing unit (5) to which the matching condition is imposed; High-speed imaging method.

9. 9. The high-speed imaging method according to claim 7 or 8, The light-sensing units (5) in the array are arranged in a substantially rectangular matrix; The light-sensing unit (5) is adjacent to another light-sensing unit (5), the light-sensing unit (5) having a matrix position (i, j), where i is a row in the matrix and j is a column in the matrix, and the other light-sensing unit (5) having a matrix position (i, j±1) and optionally (i±1, j) or (i±1, j±1), High-speed imaging method.

10. 9. The high-speed imaging method according to claim 1, further comprising the steps of: a) monitoring, by an evaluation circuit for each of the light-sensing units (5), the detection signals of the light-sensing units (5) over a calibration period in at least two, preferably at least three, calibration windows in the calibration period, the calibration windows being separated in time from one another, and during the calibration period the light-sensing units (5) are not actively illuminated by one or more light sources; b) classifying said light-sensing units (5) based on the number of calibration windows observed by said light-sensing units (5); performing a calibration routine having High-speed imaging method.

11. 11. A method for high-speed imaging according to claim 10, wherein the light-sensing units (5) are divided into at least two, preferably three, classes, imposing a duration condition related to the classification of said light-sensing unit (5) in order to generate a confirmation signal for said light-sensing unit (5); High-speed imaging method.

12. The high-speed imaging method according to claim 1, wherein the light-sensing unit is a single-photon detector SPD, preferably a single-photon avalanche detector SPAD. High-speed imaging method.

13. A detector element (6) for detecting electromagnetic radiation, comprising: a. a light-sensing unit (5), preferably a single-photon detector SPD (5), more preferably a single-photon avalanche detector SPAD (5), configured to generate a signal C_ov describing the presence or absence of observed electromagnetic radiation to be detected during successive observation windows (2); b. a memory element (7) for storing the state of said light-sensing unit (5) at the end of at least N previous observation windows (2), where N is a natural number greater than or equal to 1; c) a logic circuit (4) configured to generate a confirmation signal if the signal of the light-sensing unit at the end of a current observation window (2) and the signals of the light-sensing unit in at least M observation windows (2) of the N previous observation windows (2), where M is a natural number greater than or equal to 1 and at most equal to N, indicate observation by the light-sensing unit (5); having Detector element (6).

14. In a high-speed imaging sensor system, a. a matrix / array of sensing elements (6) according to claim 13; b. A circuit (4) for reading the detector elements (6) in said matrix / array; having High-speed imaging sensor system.

15. 1. A high-speed imaging sensor system, comprising: a. a matrix / array of detector elements (6) comprising optical sensing units (5), preferably single photon detectors (SPDs) (5), more preferably single photon avalanche detectors (SPADs) (5), configured to generate a signal C_ov describing the presence or absence of observed electromagnetic radiation (photons) to be detected during successive observation windows (2); b. A logic circuit (4) that reads the signals C_ov of the detector elements (6) in said matrix / array and that has one or more memory elements (7); and the one or more memory elements (7) are configured to store states of the light-sensing units (5) at the end of at least N previous observation windows (2), where N is a natural number greater than or equal to 1; the logic circuit (4) is configured to generate a confirmation signal if the signal of the light-sensing unit (5) at the end of a current observation window and the signals of the light-sensing unit (5) in at least M observation windows (2) of the N previous observation windows (2) indicate an observation by the light-sensing unit (5), where M is a natural number greater than or equal to 1 and at most equal to N; High-speed imaging sensor system.

16. 16. The high-speed imaging sensor system according to claim 14 or 15, The high-speed imaging sensor system includes: configured to carry out the high-speed imaging method according to any one of claims 1 to 8, High-speed imaging sensor system.

Images