Optoelectronic device and method for distance measurement

Abstract

The application relates to an optoelectronic device and a method for distance measurement. The optoelectronic device (10) has an emitting device (12) for emitting an optical signal (16) which is modulated at at least one first modulation frequency f1, a receiving device (26) with at least one optical receiving element (26a) for generating a receiving signal from a reflected optical signal (22) in a detection region (18), and a control and evaluation unit (28) which is designed to determine a phase offset and an optical flight time between the emitted optical signal and the reflected optical signal and to evaluate a first amplitude determined by the receiving signal in order to identify an erroneous measurement as a result of a reflection of the optical signal at an object (20') in the detection region outside a clear region of a phase-based optical flight time method by changing the first modulation frequency f1 in a first fluctuation range during an optical flight time measurement and evaluating a first amplitude of the receiving signal reduced thereby.

Description

[0001] The present invention relates to an optoelectronic device and method for measuring distance in a detection area using a phase-based time-of-flight method according to claims 1 and 13.

[0002] In time-of-flight (TOF) distance measurement, a light signal is emitted and then received again after being reflected by an object. The time of flight is proportional to the distance to the object and can be calculated based on the speed of light. This type of distance measurement is also known as TOF or LIDAR (Light Detection and Ranging). Sensors that utilize this principle include rangefinders and laser scanners. In this way, 3D time-of-flight cameras obtain depth information using their individual pixels. The detected three-dimensional image data with distance or spacing values ​​for each pixel is also called a 3D image, distance image, or depth map. Highly integrated TOF image sensors can be used for this purpose.

[0003] The additional distance dimension can be used in a variety of applications to obtain more information about objects in a scene detected by the camera, thereby solving a variety of tasks. For example, in automation technology, objects can be detected and classified based on 3D image information to allow for further automated processing steps based on which objects are identified (preferably including their position and orientation). Thus, for example, it can support the control of robots or different types of actuators at conveyor belts.

[0004] For vehicles operating on public roads or in enclosed environments, particularly in factory and logistics automation, 3D cameras should be used to comprehensively and three-dimensionally survey the entire environment, especially the planned routes. This applies to virtually any conceivable vehicle, whether driver-operated (PKW), LKW, work machines, and forklifts, or driverless (AGV), AGC, AMR, or ground transport vehicles. The image data is used to enable autonomous navigation or to support drivers in identifying obstacles, avoiding collisions, or facilitating the loading and unloading of transported goods, including cartons, pallets, containers, or trailers.

[0005] In safety technology, 3D cameras detect people around hazardous points (e.g., machines, especially robots). If an unacceptable intrusion into a protected area or below a safe distance from the machine is detected, a safety-oriented response is initiated to prevent accidents. Sensors used in safety technology must operate with exceptional reliability and therefore must meet high safety requirements, such as the standard EN 13849 concerning machine safety and the equipment standard EN 61496 concerning non-contact protective systems (BMS). Meeting these safety standards requires a range of measures, such as electronic assessments of safety through redundant and diverse electronics, functional monitoring, or specialized monitoring for contamination of optical components. Safety standards also specify so-called safety levels or safety requirement grades to classify achievable safety levels. Examples include SIL (Safety Integrity Level) conforming to IEC 61508 / IEC 61511, where 1 is the lowest level and 4 is the highest, or PL (Performance Level) conforming to EN ISO 13849-1, where a is “low” to e is “high”.

[0006] In phase-based time-of-flight methods (also known as CW measurement methods or iTOF (indirect TOF)), the scene is illuminated with amplitude-modulated light at a modulation frequency typically on the order of 100 MHz. Light returning from the scene is received and demodulated at the same frequency (Lock-In-Verfahren), the same frequency used to modulate the transmitted light. Demodulation yields an amplitude measurement corresponding to a sampled value of the received signal. According to the Nyquist criterion, at least two samples are required to determine the phase of the periodic signal. Therefore, the measurement is performed with a difference in relative phase between the signals modulated on the transmitting side and demodulated on the receiving side. This allows the determination of the absolute phase shift between the transmitted and received signals, which is proportional to the distance to objects in the scene.

[0007] However, the phase method has only a limited defined region corresponding to the period of the modulation frequency. Due to the dual optical path, this equates to an effective range of only 1.5 m at a modulation frequency of 100 MHz. To increase the defined region and thus the effective range, multiple different modulation frequencies are typically used for measurement. Alternative reductions in the modulation frequency are only considered within a narrow range, as this would affect measurement accuracy. Multi-modulation frequency measurement combines higher depth resolution with a larger depth measurement range.

[0008] If an object is located at a distance outside the defined region denoted by D, it is, in principle, impossible to determine whether it is associated with the region [0, D), [D, 2D), ... This applies to simple defined regions measured at a single modulation frequency, as well as to defined regions measured at a combination of multiple modulation frequencies. Therefore, erroneous measurements may occur when the object is located at a large distance beyond the practically effective range of the first defined region [0, D). Simple intensity filters coordinated with quadratic distance reduction (Entfernungsabnahme) cannot reliably filter out such objects, especially when the object is very bright, glossy, or even has a back-reflective surface. Then, despite the large distance, correlated and confusing signal strengths remain. Most importantly, this indistinguishability raises questions about the use of security technologies, as incorrectly associating measurements of distant objects with nearby regions can severely impair the functionality of equipment or render it unusable in security-related situations.

[0009] In theory, a very large defined region can be achieved by appropriately selecting the modulation frequency. However, in real-world, noisy signals, this does not always lead to useful, accurate measurements. US2014 / 0049767A1 discloses a method for geometric phase unwrapping of optical time-of-flight systems that, instead of aiming for the largest possible defined range, cites two frequency examples of 31 MHz and 50 MHz to address the challenges from its perspective. Alternatively, it specifically uses frequencies derived from a relatively large common fundamental frequency a, specifically three frequencies f1 = am1, f2 = am2, and f3 = am3, where m1, m2, and m3 are small, coprime integers. By using this method, although measurements are performed with multiple modulation frequencies, the defined region is reduced to, for example, on the order of ten or twenty meters, where objects at distances outside the defined region may have considerable practical relevance. However, US2014 / 0049767A1 does not provide a solution for this.

[0010] EP 3 736 601 A1 relates to determining the distance to an object while taking edge impacts into account. Here, distance is measured based on phase using two frequencies. Furthermore, the amplitude of each frequency is evaluated, as it differs from the amplitude of an impact on the entire cross-section of an object at the same distance when the scanning beam separately impacts a portion of the cross-section at different distances. However, aside from mentioning common methods with multiple modulation frequencies, this document makes no contribution to defining specific regions.

[0011] US2018 / 0210070 A1 describes an active, brightness-based strategy for finding invalid pixels in an optical time-of-flight system. This addresses multipath propagation and motion blur. Similar to EP 3 736 601 A1, this evaluates amplitudes, and if these amplitudes can infer a mixture of detected distances, the pixel is set as "invalid." Again, the issue of received signals from objects outside the defined region is not discussed, and only the possibility of increasing the defined region through multiple modulation frequencies is briefly mentioned.

[0012] Therefore, the objective of this invention is to improve distance measurement using a phase-based time-of-flight method.

[0013] This task is accomplished by the optoelectronic device according to claim 1, particularly a 3D time-of-flight camera, and the method according to claim 13 for measuring distance in a detection area using a phase-based time-of-flight method. Time-of-flight measurement works essentially as described at the outset. A transmitting device generates a light signal modulated with a first modulation frequency f1. As a precaution, this should be considered an artificial amplitude modulation, and the chosen modulation frequency should not be confused with the carrier frequency of the actual light wave. The emitted light reflected from an object in the detection area, superimposed with incoming or background light, is incident on a receiving device having at least one light-receiving element, which thereby generates a received signal. Here, particularly relevant is an image sensor having multiple light-receiving elements or pixels for detecting three-dimensional image data, each light-receiving element or pixel generating its own received signal for that image point. In the reflected light, there is no conceptual distinction between directional reflection and non-directional diffuse reflection or scattering.

[0014] The control and evaluation unit uses a phase method to determine the time of flight of light and thus the distance value r, where explicit conversion from the speed of light to metric data is not absolutely necessary. For phase determination, the received signal is demodulated at a first modulation frequency f1 already used for the emitted light signal using a locking method. Demodulation can be implemented already in the light-receiving element of the image sensor or in a pixel, also known as a TOF pixel (Time of Flight) or locked pixel. Other functions of the control and evaluation unit can also be integrated into the receiving device, particularly on a common chip. Alternative further processing can also be available downstream of the receiving device, for example, in an FPGA (Field-Programmable Gate Array).

[0015] In addition to measuring the time of flight of light, amplitude evaluation is performed to detect measurement errors. The measured distance may be erroneous for various reasons, or more precisely, not in the sense of measurement error, but entirely different from the true distance. Therefore, at least a first amplitude at at least a first modulation frequency f1 is determined and evaluated from the received signal.

[0016] This invention is based on the fundamental idea of ​​identifying and preferably correcting specific erroneous measurements, namely, whether the distance to an object outside a first defined region [0, D). More formally, an object may be located in reality at a distance r+nD, n>1, and may be erroneously measured as being particularly close at a distance r. This is a completely different type of erroneous measurement from multipath propagation or ghosting, requiring special measures to detect.

[0017] During optical time-of-flight measurements, instead of using a fixed modulation frequency f1, or, in the embodiments described below, multiple fixed modulation frequencies f1, f2, ..., are used, the modulation frequency itself is varied. This can be called frequency modulation, but due to the doubling of the modulation frequency in frequency modulation, this term will be avoided hereinafter and referred to as spread spectrum. In any case, the modulation frequency is varied, diffuse, or jittery, and this is reflected in the reduced initial amplitude for reasons explained later. Spread spectrum also cannot be used with multiple modulation frequencies f1, f2, ... i Multiple measurements are confused, which involves rapid changes within the same phase measurement, using a further modulation frequency f. i Subsequent measurements can be modified again with the same spread spectrum, a different spread spectrum, or no spread spectrum.

[0018] The reduction in at least the first amplitude caused by spread spectrum is assessed. The effect of spread spectrum on the first amplitude can be calculated, thus an expected value for the first amplitude exists in a first defined region [0, D) or in a more distant region [D, 2D), [2D, 3D), ... . From this expected value, a constant or distance-dependent amplitude standard can be derived, and the measured optical time of flight is assigned to the first defined region [0, D) or the more distant region [D, 2D), [D, 3D), ... by comparison with the measured first amplitude or the value derived therefrom.

[0019] The advantage of this invention lies in its reliable identification of erroneous measurements caused by objects outside the first defined region. Specifically, it distinguishes between objects within the first defined region [0, D) and objects within subsequent defined regions [D, 2D), [2D, 3D), ... . This requires no fundamental changes to the measurement principle, only an extension of the modulation from pure amplitude modulation to frequency modulation or spread spectrum. Therefore, the resulting decrease in measured amplitude is not dependent on interference effects of the measurement conditions, as in the case of multipath propagation or ghosting. Instead, it generates additional distance information in the amplitude in a completely targeted and reproducible manner. Even under ideal measurement conditions, the problem of defined regions arises as erroneous measurements in a different sense than multipath propagation or ghosting, and corresponds to inherent problems in the measurement principle, where amplitude information, which is also related to ideal conditions, is applied and evaluated. Of course, this does not mean that robustly evaluating amplitude information is impossible or unreasonable, and that the correct defined region can be found even under real, disturbed measurement conditions.

[0020] Furthermore, the erroneous measurement discovered according to the invention is only erroneous in the first step. Similarly, compared to multipath propagation or ghosting, it is possible to determine what kind of error has occurred, so such measurements need not be discarded or pixels marked as invalid; instead, new, corrected distance values ​​can be assigned. Spread spectrum improves electromagnetic compatibility (EMC) and coexistence with other similar or structurally identical systems. This is not the primary objective in the context of this invention, though synergistic advantages are nonetheless present.

[0021] Preferably, the control and evaluation unit is designed to change at least a first modulation frequency f1 within a first fluctuation range using a monotonic function, particularly a ramp or sine curve. One way to formally specify spread spectrum is a function Δf(t), which is added to the modulation frequency. Preferably, this function is monotonic, at least for a certain time interval, so as to traverse the fluctuation range once or multiple times. An example of this is a ramp or sine curve that traverses the fluctuation range once or multiple times from bottom to top, from top to bottom, or back and forth. Preferably, the function is unbiased or symmetrical about zero so as not to shift the average or effective modulation frequency, which would otherwise be considered in the time-of-flight determination.

[0022] Preferably, the control and evaluation unit is designed to modulate the optical signal at at least one second modulation frequency f2 and receive the optical signal again. By performing measurements at multiple modulation frequencies, the defined region can be expanded and / or the measurement accuracy improved, as already explained in the introduction. Preferably, the frequencies can be selected according to the teachings of US2014 / 0049767 A1 cited in the introduction, i.e., there exists a common fundamental frequency a, typically in the range of several MHz to twenty MHz, and multiple modulation frequencies are multiples of it am1, am2, ..., where preferably m1, m2, ... are pairs of coprime integers. Utilizing the phase unpacking proposed in US 2014 / 0049767 A1 does not necessarily expand the defined region. For example, it can be envisioned that the lowest frequency contributes to a relatively large defined region, while the remaining frequencies, especially the highest frequency, contribute to higher measurement accuracy.

[0023] Preferably, the control and evaluation unit is designed to determine and evaluate at least one second amplitude from the received signal at at least the second modulation frequency f2. Therefore, when measuring using multiple modulation frequencies, it is preferable to determine and evaluate the relevant amplitudes separately in order to find the correct, well-defined region.

[0024] Preferably, the control and evaluation unit is designed to change the second modulation frequency f2 within a second fluctuation range different from the first fluctuation range. Therefore, different spread spectrums are used for multiple modulation frequencies. At more than two modulation frequencies, the spread spectrum can be repeated, or each modulation frequency can have its own spread spectrum. Alternatively, the same spread spectrum can be used for all modulation frequencies. Spread spectrum can be understood as absolute or relative. Therefore, the function Δf(t) is fixedly given beforehand, or given beforehand, for example, a relative fluctuation range of a certain percentage of the corresponding modulation frequency, and the curve of Δf(t) adapts to that fluctuation range. It is conceivable that at least one modulation frequency does not undergo spread spectrum. In practice, the difference can be considered from the amplitude variation caused by spread spectrum. This could also, of course, mean the same percentage fluctuation range that has different effects due to different associated modulation frequencies.

[0025] Preferably, the control and evaluation unit is designed to evaluate the first amplitude based on a distance-dependent amplitude loss function h(r) of the ratio of the amplitude at a changed modulation frequency to the amplitude at a constant modulation frequency, specifically based on each modulation frequency f. i Distance-dependent amplitude loss function h i (r) is used to evaluate the i-th amplitude. If the amplitude during spread spectrum is represented by I... int If the amplitude without spread spectrum is denoted by I, then h(r) = I is given. int (r) / I(r) represents the normalized, distance-dependent amplitude loss through spread spectrum. h(r) is calculated from a modulation frequency f1 or multiple modulation frequencies f...i h i (r) can derive criteria, particularly thresholds, for determining the precise region in which the measured distance falls. The amplitude loss function is the expected value, and the i-th amplitude is the corresponding measured value, which is checked against this expected value to find a suitable precise region. First, the amplitude loss function is applied to different modulation frequencies f. i Individual checks are conducted (in the case of multiple modulation frequencies), followed by a joint decision.

[0026] Preferably, the control and evaluation unit is designed to use combined rule cancellation (miteiander zuverrechnen) to cancel multiple modulation frequencies f i The i-th amplitude, and based on the amplitude evaluation function k(r) combined according to the combination rule for all used modulation frequencies f. i An evaluation is performed. Preferably, the amplitude evaluation function k(r) combines the expected values ​​of multiple modulation frequencies used, wherein, as already done in h(r), the amplitude evaluation function k(r) itself need not be the expected value, but rather a threshold can be derived from it, for example. Therefore, the i-th amplitudes measured separately are combined with each other and thus compared with the expected value in the step. The combination rule links multiple input variables as in the case of summation, thereby immediately proposing a suitable combination rule. This can be applied to the amplitude loss function h. i (r) can also be applied to the i-th amplitude being measured.

[0027] Preferably, the combination rule is a ratio or quotient. This has the advantage that variables such as intensity curves and diffuse reflectance, which are purely distance-dependent and typically exhibit quadratic decreases, are simplified. Generally, any function, preferably a linear combination or a single value, is possible in both the numerator and denominator. Particularly preferably, the combination rule is a dispersion measure. The ratio or quotient of the standardization measure. An exemplary dispersion measure is the difference or standard deviation between the maximum and minimum values, or, for simpler calculation, the sum of squares corresponding to the variance. Exemplary standardization measures are the maximum, sum, or mean.

[0028] Preferably, the combination rule considers only the largest and / or smallest modulation frequencies. Unlike the incalculable measurement errors caused, for example, by multipath propagation or ghosting, the amplitude loss introduced by spread spectrum is known and predictable. Therefore, it is particularly unnecessary to dynamically determine the maximum or minimum values. Instead, it is known from the outset which modulation frequencies suffer the largest and smallest amplitude losses due to spread spectrum, and this is invariant, at least over large distances covering a clearly defined area. This should be taken into account if spread spectrum with the same fluctuation width or producing the same amplitude loss is not used for all modulation frequencies, as this could relativize or even invert the ratio. In any case, due to the known and reproducible amplitude loss, the combination rule can be simplified by using only variables specifically for particular modulation frequencies, particularly without considering one or more modulation frequencies in this respect.

[0029] Preferably, the control and evaluation unit is designed to evaluate amplitude based on a threshold. This is a very simple and easily verifiable standard. Preferably, it is based on an amplitude loss function h. i The threshold is found using the amplitude evaluation function k(r) and specifically, the first defined region [0,D) is separated from the more distant defined regions [D,2D)...

[0030] Preferably, the threshold is a function of distance. At this point, using this formula eliminates mathematically conceivable constant functions, thus allowing for more precise adjustment of the distance-related threshold and enabling better differentiation.

[0031] Preferably, the control and evaluation unit is designed to correct erroneous measurements of the measured distance by a multiple of the defined region. If the object is determined to be at a distance beyond the first defined region [0, D) using the method according to the invention, it is preferable not to discard the measurement or mark the relevant pixels as invalid. Instead, the measurement is corrected by an extension D of the defined region, i.e., the new distance is set as r+nD. In principle, the i-th amplitude of the measurement can even determine n>1, so the correct multiple of the extension D of the defined region can be added. However, in a preferred embodiment, it is simply added to D where n=1, i.e., the measurement is set as r+D. This is the most likely case because even distant objects are almost undetectable due to the decrease in quadratic intensity associated with distance, even with very high diffuse reflectivity. In any case, in security applications, the conservative assumption is to provide sufficient opportunity to measure the object again at a later time point when the movement approaches, at the latest when the object reaches the first defined region. For high-precision quantitative measurements over a large effective range, the defined range itself should be extended in any case, for example, by using additional modulation frequencies. If the defined area is at least practically doubled through this invention, then that is already a great achievement.

[0032] The method according to the invention can be further developed in a similar manner and exhibits similar advantages. These advantageous features are described, exemplarily but not exhaustively, in the dependent claims which are subordinate to the independent claims. Attached Figure Description

[0033] Other features and advantages of the invention will now be described in more detail, based on exemplary embodiments and with reference to the accompanying drawings. In the drawings:

[0034] Figure 1 A schematic block diagram of a 3D optical time-of-flight camera is shown.

[0035] Figure 2 A diagram showing distance measurements of objects within and outside a first defined area is provided.

[0036] Figure 3 An exemplary depth map measured using a 3D optical time-of-flight camera is shown, which has erroneous measurements due to ambiguity;

[0037] Figure 4 An exemplary spread spectrum is shown;

[0038] Figures 5a-5c A vector diagram illustrating the effect of different degrees of spread spectrum on amplitude is shown;

[0039] Figure 6 An exemplary diagram of the spread spectrum-based amplitude loss function h(r) is shown;

[0040] Figure 7 Multiple modulation frequencies f are shown i Multiple amplitude loss functions h i An exemplary diagram of (r);

[0041] Figure 8 It shows the relationship with the data. Figure 7 A graph showing the sum of the amplitude loss function h(r) and the combined amplitude evaluation function k(r);

[0042] Figure 9 It shows that according to Figure 8 The plot of the magnitude evaluation function k(r) is now a cyclic plot in multiple defined regions and has a constant threshold in the first defined region;

[0043] Figure 10 This shows a similar threshold that is now related to distance. Figure 9 The illustration; and

[0044] Figure 11 It shows that according to Figure 3 An example depth map after correcting for ambiguity.

[0045] Figure 1 A schematic block diagram of a camera 10 designed as a 3D time-of-flight camera is shown. The invention is illustrated using camera 10 as an example, but generally, for phase-based time-of-flight methods, the invention can also be used with other sensors, such as rangefinder scanners, laser scanners with rotating mirrors or measuring heads, or solid-state laser scanners.

[0046] The illumination unit 12 emits emitted light 16, modulated by the emitting optics 14, into the detection area 18. An edge emitter or an LED or laser in the form of a VCSEL can be considered as the light source. The illumination unit 12 is operable such that the amplitude of the emitted light 16 is modulated at a frequency typically in the range of 1 MHz to 1000 MHz. For example, the modulation can be sinusoidal or rectangular, and in any case, periodic.

[0047] If the emitted light 16 strikes the object 20 in the detection area 18, a portion is reflected back to the camera 10 as received light 22, and there it is guided to the image sensor 26 by receiving optics 24 (e.g., a single lens or receiving objective). The image sensor 26 has multiple receiving elements or receiving pixels 26a. The resolution of the image sensor 26 can range from two or a few receiving pixels 26a to thousands or millions of receiving pixels 26a. The pixel arrangement is typically a matrix, thus obtaining lateral positional resolution along the X and Y directions, supplemented by distance measurement in the Z direction, forming three-dimensional image data. When referring to 3D cameras, 3D time-of-flight cameras, or three-dimensional image data, 3D detection is preferred. However, other pixel arrangements can also be considered in principle, such as pixels selected in a matrix or forming an entire row of pixels in the image sensor of a line scan camera.

[0048] Each receiving pixel 26a includes a photosensitive surface and at least one charge memory, such as a capacitor. Here, in the locking method, the timing of collecting the charge generated by the incident receiving light 22 into the corresponding charge memory is controlled by using a demodulation frequency corresponding to the modulation frequency used for emitting light 16 during the measurement duration or integration time.

[0049] The charge quantity in the charge memory of receiving pixel 26a is read out, digitized, and transmitted to control and evaluation unit 28. To obtain additional sampled values ​​for phase determination, the described partial measurement, along with the collection and readout of the charge memory, is repeated n times, preferably 2 to 4 times. In each case, the phase between the modulation frequency used for emitting light 16 and the demodulation frequency used for the locking method is different. Multiple charge memories can replace this partial measurement in a certain way, since these charge memories have already generated multiple sampled values ​​in the same partial measurement, or these charge memories can be used for differential measurements to compensate for certain asymmetries in the pixel structure. Preferred constellation positions (Konstellation) for the sampled values ​​are measurements with phases of 0°, 90°, 180°, and 270° or 0°, 120°, and 240°.

[0050] Now, the control and evaluation unit 28 reconstructs the phase shift from multiple sampled values ​​based on the time-of-flight of light passing through the detection region 18, which can be converted into a distance value for each receiving pixel 26a. A three-dimensional image, distance image, or depth map is generated and output at interface 30. Interface 30, or alternatively one or more other interfaces not shown, is used in turn to input control signals to camera 10 or to parameterize camera 10.

[0051] Will Figure 1 The division into an image sensor 26 with receiving pixels 26a (each receiving pixel demodulates itself) and a control and evaluation unit 28 is only a preferred embodiment. This enables a highly integrated image sensor 26, where at least a portion of the modulation signal, demodulation signal, and other measurement techniques for the illumination unit 12 are integrated into the image sensor. However, the control and evaluation functions can also be distributed in other ways, for example, the control and evaluation unit 28 can handle at least a portion of these tasks of the image sensor 26, which would be simpler. The control and evaluation unit 28 does not necessarily have to be a single component as shown, but can be composed of one or more digital computing components, such as a microprocessor, FPGA (Field-Programmable Gate Array), or ASIC (Application-Specific Integrated Circuit). Furthermore, the illumination shown is planar illumination, for which a diffuser is used, for example, as part of the emitting optics 14. In another embodiment, the arrangement of multiple individual light sources of the illumination unit 12 is clearly projected into the detection area 18, so that the receiving pixels 26a are individually illuminated and the effective range is increased. In addition, unlike the illustration, the lighting may not be integrated into the camera 10, but may be structurally or spatially separate from the camera 10.

[0052] Figure 2Distance measurements are shown for an object 20 within a first unambiguity range and for an object 20' outside the first unambiguity range (e.g., within a second unambiguity range). In the distance measurements, the phase shift between the emitted light 16 generated by the time-of-flight of light and the received light 22 reflected by the objects 20 and 20' is illustrated. The following is determined:

[0053]

[0054] Wherein, at the modulation frequency f and the speed of light c, D = c / (2f).

[0055] However, due to the 2π periodicity of the modulation frequency, the two objects 20 and 20' produce the same phase shift even with practically different object spacings r and r+D. Generally, for r = d + n*D, n = 0, 1, ..., we obtain:

[0056]

[0057] Therefore, phase measurements only have a finite defined region D (or a first defined region [0, D), a second defined region [D, 2D), and other defined regions. At this point, the language is not always perfectly clear, and sometimes an undefined defined region is used to represent the first defined region [0, D). The defined region can be increased by choosing a small modulation frequency f, but at the cost of low measurement resolution. Therefore, it is preferable to perform measurements at two to three or more modulation frequencies to increase the defined region and / or obtain high measurement accuracy in the combination of measurements. This has already been discussed in the introduction. However, even with these measures, a finite defined region still exists, and thus essentially continues as follows: Figure 2 The situation shown in the figure.

[0058] Typically, distance measurements are associated with a first defined region [0, D) when n = 0. Distant objects 20' usually have only small signal strength I and are either not detected at all or can be picked up by a noise threshold. However, especially in the case of objects 20' with high diffuse reflectivity (glossy surfaces, reflectors, or even retroreflectors), sufficient signal strength is maintained despite the greater distance. Object 20' is then incorrectly measured as distance r instead of the actual distance r + D. This can lead to errors in scene evaluation and, in security applications, critical errors or at least unnecessary security-oriented measures. Higher-order errors, i.e., errors when n > 1, can also be considered, but here the distance-related intensity drop usually dominates.

[0059] The ambiguity described is an inherent characteristic of phase measurement methods and is conventionally accepted. Alternatively, the described method is used to define the region through small and / or multiple modulation frequency spreads. However, this does not preclude the detection of object 20' outside the obtained first defined region. The method according to the invention is described below to associate objects 20, 20' with the correct defined region, or at least determine whether objects 20, 20' are actually located within the first defined region.

[0060] Figure 3 First, an exemplary false-color or grayscale encoded depth map without correction according to the invention is shown. In the background space of the scene, the circled area marked with an exclamation mark is identifiable, and ambiguity in this area leads to misclassification in the near-field region. This gives incorrect significance to actually distant objects in scene evaluation. In security technology analysis utilizing protective fields, this could potentially lead to security-oriented cut-offs or other unnecessary protective measures, and thus usability issues.

[0061] Figure 4 An exemplary spread spectrum approach introduced to address a specific problem is illustrated. Here, spread spectrum means that instead of modulating the emitted light 16 at a fixed frequency f, the illumination unit 12 selectively alters, modulates, or dithers that frequency over time. Typically, this approach is sometimes used to improve coexistence with other systems or to achieve higher EMC compatibility by reducing radiation at a specific frequency. The present invention can also utilize these advantages, but these are not the effects discussed here. A conceivable mathematical formula for spread spectrum is f(t) = f + Δf(t), where Δf(t) is the additional spread spectrum. Figure 4 The specific spread spectrum curve in the example is illustrative.

[0062] Figures 5a-5c This explains the function of spread spectrum. Here, Figure 5a This shows the case without spread spectrum. Figure 5b This illustrates the case of moderate spread spectrum, and Figure 5c This illustrates a strong spread spectrum scenario. The light-colored arrows and gray background represent transient signals. The black arrow indicates the time integral above it, thus representing the amplitude I measured for this purpose. int Spread spectrum causes an instantaneous phase change in the received signal. The time modulation, more specifically: in:

[0063]

[0064] Therefore, phase modulation It is proportional to the frequency offset Δf(t) and the distance r of the object.

[0065] In Figures 5a-5c it, the influence on the instantaneous phase and the measurement is shown in vector diagram, that is, the integration of the received signal within the integration time. It can be seen that compared with the intensity I without spreading spectrum as in Figure 5a the intensity I of the integrated signal int is reduced: I int < I. Here, the degree of amplitude loss depends on the spreading spectrum and also depends on the distance as will be further elaborated. Therefore, when the spreading spectrum is known, an additional distance correlation is imposed on the amplitude, and this distance correlation can be used to resolve the periodic ambiguity of the phase measurement or the association in the clear area. The terms intensity and amplitude are used interchangeably in many places in this specification.

[0066] In order for the spreading spectrum not to change the further evaluation of the fundamental modulation frequency, the unmodified modulation frequency f, the phase of the signal averaged within the integration time should not change, that is should apply. This can be achieved by a balanced or symmetric spreading spectrum Δf(t) within the integration time. Examples of suitable spreading spectra Δf(t) are ramps or sine curves that cross the fluctuation range opened by the spreading spectrum one or more times in one or two directions within the integration time. In Figure 4 a rising ramp and a falling ramp in two directions are shown as one of these examples.

[0067] Figure 6 shows an exemplary curve of the normalized amplitude loss h(r) due to the spreading spectrum related to the distance r (in terms of the clear area D). In the case of no spreading spectrum and in the idealized case without distortion effects such as those caused by multipath propagation or ghosting, the intensity or amplitude related to the distance is expected to be as follows:

[0068] I(r) = R * g(r) * A

[0069] Here, R is the diffuse reflectivity of the detected object 20, g(r) is the intensity loss related to the distance, usually and A represents the remaining correlation, such as the intensity of the emitted light 16 of the radiation, the modulation efficiency, etc., and should not be understood as the amplitude, and the amplitude has been represented by the intensity I.

[0070] With the spreading spectrum, the integrated intensity I int also has an additional distance-related amplitude loss:

[0071] I int (r) = R * g(r) * h(r) * A.

[0072] The additional distance-related factor is in Figure 6 The amplitude loss function h(r) = I shown in the figure int (r) / I(r). It only gives the adjusted and normalized distance-dependent effect of spread spectrum on amplitude or intensity.

[0073] Figure 7 Multiple modulation frequencies f are shown i Multiple standardized amplitude loss functions h i The diagram for (r) illustrates this. As mentioned several times, multiple different measurement frequencies f1, f2, ..., each with its own defined region D1, D2, ..., can be used to jointly evaluate the phase of the measurement. Increasing the resulting definite region D, i.e., D = D1m1 = D2m2 = ..., where m1, m2, ... are integers. However, a fundamental problem remains with the new definite region D. Different objects 20, 20' that are a multiple of D in distance produce the same phase, i.e., for r = d + n * D, n = 0, 1, ...:

[0074]

[0075] According to the present invention, the above-described spreading Δf(t) is applied to modulation frequencies f1, f2, ... . This applies to each modulation frequency f i Individual options are available. For example, the same spread spectrum can be applied to all modulation frequencies f. i Or different spread spectrums can be applied to each modulation frequency f. i Alternatively, the same and different spread spectrums can be applied in any mixed manner to the modulation frequency f. i It is also conceivable to not change at least one or exactly one modulation frequency f. i Or only at the modulation frequency f i Spread spectrum is then performed. However, without spread spectrum, for the relevant modulation frequency f... i h i ≡1, therefore it makes no contribution to the association with a defined region, or can only be used as a reference in the quotient (Quotientenbildung) further discussed below. Therefore, not all modulation frequencies f i Neither of them has spread spectrum, so there will be no amplitude loss that can be used for association with a specific region.

[0076] Similar to a modulation frequency f alone Figure 6 The explanation now applies to multiple modulation frequencies f without spread spectrum. i :

[0077] I1(r)=R*g(r)*A1

[0078] I2(r)=R*g(r)*A2

[0079] …

[0080] Spread spectrum is now used for each modulation frequency f i An amplitude loss factor h related to distance was added. i (r):

[0081] I int,1 (r)=R*g(r)*h1(r)*A1

[0082] I int,2 (r)=R*g(r)*h2(r)*A2

[0083] …

[0084] Figure 7 Example values ​​h1(r), h2(r), and h3(r) for three modulation frequencies f1, f2, and f3 are shown. Exemplary values ​​are f1 = 120 MHz, f2 = 80 MHz, and f3 = 16 MHz. The defined region D is 18.75 m corresponding to the three selected frequencies. Modulation frequency f i A smaller common divisor can be used to select the region, resulting in a larger definite region D, doubled compared to the slowest frequency f3 in the digital example. Instead of maximizing the definite region, in this example, multiple modulation frequencies f... i Primarily used for higher measurement accuracy and only partially for increased defined regions. This compensation between defined region and measurement accuracy can also be achieved in different ways. For this purpose, the measurement frequency can be advantageously selected and relevant phase unpacking can be used, as in US 2014 / 0049767A1. Regardless of how the defined region D is ultimately generated, the problem of periodic ambiguity will always arise.

[0085] Figure 8 It shows Figure 7 amplitude loss function h i (r) is the superposition of the combined amplitude evaluation function k(r). In principle, it can be envisioned that for each modulation frequency f... i The amplitudes measured separately are evaluated individually, for example, based on the amplitude loss function h. i (r) The corresponding standard is derived. However, the dependence on diffuse reflection behavior and other variables still exists. This is while considering multiple or all modulation frequencies f. i The magnitude evaluation function k(r) for merging is easier to handle.

[0086] The starting point is already... Figure 7 Each introduced modulation frequency f i Distance-related integral strength:

[0087] I int,1(r)=R*g(r)*h1(r)

[0088] I int,2 (r)=R*g(r)*h2(r)

[0089] …

[0090] To simplify further explanation and evaluation, we assume here that A1 = 1, A2 = 1, ... This can also be achieved in practice through appropriate calibration measurements.

[0091] If these integral intensities are correlated in an appropriate combination, the common terms R and g(r) are simplified, and what remains is a magnitude evaluation function k(r) given by h1(r), h2(r), ... which is purely distance-dependent. Figure 8 The example shown is of three measured frequencies f1, f2, f3:

[0092]

[0093] This is by no means the only conceivable amplitude evaluation function k(r). Preferably, other amplitude evaluation functions k(r) are similar, as they also associate a dispersion measure (Streumaβ) with a standardized measure. Here, the numerator may include a sum of squares or standard deviation, and the denominator may include a sum or maximum value. Using both dispersion and standardized measures is advantageous, but not necessary. Preferably, at least one quotient should be formed to offset the effect of R*g(r). However, in other embodiments, the numerator and denominator may contain other linear combinations or only individual intensities.

[0094] Preferably, the intensity or amplitude actually measured in operation completely cancels each other out, i.e., the quotient is formed by a single intensity, a linear combination, or by a combination of dispersed and standardized metrics, preferably according to the same rules as the amplitude evaluation function k(r). The ambiguity problem can then be resolved by comparing the calculated result of the measured amplitude with k(r). Whether one, more, or all ambiguities can be resolved depends not only on the function k(r) but also on the achievable measurement accuracy and interference effects such as multipath propagation or ghosting.

[0095] Instead of comparing the measured amplitude or the value derived therefrom with the amplitude evaluation function itself k(r), a threshold is preferably set based on the amplitude evaluation function k(r) to simplify the evaluation. Therefore, it is possible to distinguish whether objects 20, 20' are located in the first defined region [0, D) or at a greater distance r > D.

[0096] Figure 9 An example is shown where the threshold is constant at 0.7 (corresponding to the horizontal dash). It is drawn according to... Figure 8 The representation of the magnitude evaluation function k(r). And... Figure 8 Unlike other plots, this one is cyclic (wrapped around), meaning the magnitude evaluation function k(r) extends to the right after a defined region is plotted on the X-axis, and then continues again from the left, where the magnitude evaluation function k(r) is shown as a solid line for the first defined region and as a dashed line for the other defined regions.

[0097] A constant threshold separates the first defined region from other defined regions located thereafter. The gray background area is intended to indicate that inaccurate measurements could lead to correlation problems. Therefore, in a preferred embodiment, measurements within this gray area are considered invalid.

[0098] Figure 10 Another example of a threshold correlated with distance is shown, again plotted as a dashed line, and more precisely follows the curve of the magnitude evaluation function k(r) to more accurately and selectively correlate in the first defined region or beyond. In other respects, Figure 10 Corresponding to Figure 9 For simplicity, only the indeterminate edge areas of the gray background have been omitted. Figure 9 and Figure 10 The two threshold curves in the diagram should be understood as purely exemplary, and other alternatives are conceivable, such as piecewise constant thresholds, multiple thresholds, etc.

[0099] Instead of discarding or invalidating distance measurements or pixels that do not meet the threshold criteria for the first defined region, the measured distance can be corrected by D. For this purpose, the measured value r is replaced by r+D. In principle, the actual distance could be envisioned as r+nD (n>1), but this is unlikely because distant objects 20' and 20' have very low intensity, and for many applications, knowing the distance > D is sufficient. Furthermore, if necessary, n can be determined using a more complex evaluation than a simple threshold.

[0100] Figure 11 Show again according to Figure 3 An exemplary depth map, however, is now after fuzziness correction according to the present invention. Figure 3 The erroneous correlations highlighted in the diagram are resolved through spread spectrum and the resulting assessment of distance-related amplitude loss. At most, individual erroneous pixels are retained; these pixels are inconspicuous and ignored by object segmentation or similar scene assessments. In particular, these pixels do not cause unnecessary security-oriented responses in security applications. Alternatively, individual erroneous pixels can be eliminated via filters. Therefore, high detection sensitivity, measurement accuracy, effective measurement range, and robustness are achieved in security and other applications.

[0101] Without the method according to the invention, at least complex downstream image evaluation algorithms are required to resolve erroneous associations of defined regions, for example, based on object relationships. This is not only laborious but also impossible to perform reliably in all cases without errors or simplifications. It is also conceivable to ensure that there are no objects in the scene that extend beyond the first defined region, at least no objects with high diffuse reflectance, but such a limitation on the application is undesirable and very difficult to control.

Claims

1. An optoelectronic device (10) for distance measurement in a detection area (18) using a phase-based time-of-flight method, comprising: a transmitting device (12) for transmitting an optical signal (16) modulated at at least one first modulation frequency f1; a receiving device (26) having at least one optical receiving element (26a) for generating a received signal from the optical signal (22) reflected in the detection area (18); and a control and evaluation unit (28) designed to determine the phase shift and time of flight between the transmitted optical signal (16) and the reflected optical signal (22), and to evaluate a first amplitude determined by the received signal to determine erroneous measurements. Its features are, The control and evaluation unit (28) is also designed to identify erroneous measurements caused by reflection of the optical signal (16) at an object (20') in the detection area (18) outside the defined area of ​​the phase-based time-of-flight method by changing the first modulation frequency f1 in a first fluctuation range during the time-of-flight measurement and evaluating the first amplitude of the received signal thereby reduced. The control and evaluation unit (28) is designed to evaluate the first amplitude based on a threshold; and The threshold is a function of distance.

2. The optoelectronic device (10) according to claim 1, wherein, The control and evaluation unit (28) is designed to use a monotonic function within the first fluctuation range to change the at least one first modulation frequency f1.

3. The optoelectronic device (10) according to claim 1, wherein, The control and evaluation unit (28) is designed to modulate the optical signal at at least one second modulation frequency f2 and receive the optical signal again.

4. The optoelectronic device (10) according to claim 2, wherein, The control and evaluation unit (28) is designed to modulate the optical signal at at least one second modulation frequency f2 and receive the optical signal again.

5. The optoelectronic device (10) according to claim 3, wherein, The control and evaluation unit (28) is designed to determine and evaluate at least one second amplitude from the received signal at the at least one second modulation frequency f2.

6. The optoelectronic device (10) according to claim 4, wherein, The control and evaluation unit (28) is designed to determine and evaluate at least one second amplitude from the received signal at the at least one second modulation frequency f2.

7. The optoelectronic device (10) according to any one of claims 3-6, wherein, The control and evaluation unit (28) is designed to change the second modulation frequency f2 within a second fluctuation range that is different from the first fluctuation range.

8. The optoelectronic device (10) according to any one of claims 1-6, wherein, The control and evaluation unit (28) is designed to evaluate the first amplitude based on a distance-dependent amplitude loss function h(r) that is the ratio of the amplitude at a changed modulation frequency to the amplitude at a constant modulation frequency.

9. The optoelectronic device (10) according to any one of claims 1-6, wherein, The control and evaluation unit (28) is designed to cancel multiple modulation frequencies f using a combination rule. i The i-th amplitude, and based on the amplitude evaluation function k(r) combined according to the said combination rule for all used modulation frequencies f. i An assessment will be conducted.

10. The optoelectronic device (10) according to claim 9, wherein, The combination rule is a ratio or quotient.

11. The optoelectronic device (10) according to claim 10, wherein, The combination rules only consider the largest and / or smallest modulation frequencies.

12. The optoelectronic device (10) according to any one of claims 1-6 and 10-11, wherein, The control and evaluation unit (28) is designed to correct for erroneous measurements of the measured distance by a multiple of the defined area.

13. The optoelectronic device (10) according to claim 1, wherein, The optoelectronic device (10) is a 3D optical time-of-flight camera.

14. The optoelectronic device (10) according to claim 2, wherein, The monotonic function is a slope or a sine curve.

15. The optoelectronic device (10) according to claim 8, wherein, The control and evaluation unit (28) is designed to be used based on each modulation frequency f i Distance-dependent amplitude loss function h i (r) is used to evaluate the i-th amplitude.

16. The optoelectronic device (10) according to claim 10, wherein, The ratio or quotient is a ratio or quotient of a linear combination, or a ratio or quotient of a dispersion measure and a standardized measure, wherein the dispersion measure is the difference or sum of squares between the maximum and minimum values, and the standardized measure is the sum or average.

17. A method for distance measurement in a detection region (18) using a phase-based time-of-flight optical method, wherein: The system transmits an optical signal (16) modulated at at least one first modulation frequency f1; generates a received signal from an optical signal (22) reflected in the detection region (18); determines the phase shift and time of flight between the transmitted optical signal (16) and the reflected optical signal (22); and further determines and evaluates a first amplitude from the received signal to identify measurement errors. Its features are, Errors caused by reflection of the optical signal (22) at an object (20') in a detection region (18) outside the defined region of the phase-based time-of-flight method are identified by changing the first modulation frequency f1 within a first fluctuation range during the time-of-flight measurement and evaluating the first amplitude of the received signal that is thereby reduced. Wherein, the first amplitude is evaluated based on a threshold; and The threshold is a function of distance.

18. The method according to claim 17, wherein, The method described is used for detecting three-dimensional image data.

Images