METHOD FOR ACQUIRING A DEPTH MAPPING USING INDIRECT FLIGHT TIME AND CORRESPONDING SENSOR
Patent Information
- Application Number
- DE602021022762
- Authority / Receiving Office
- DE · DE
- Patent Type
- Patents
- Current Assignee / Owner
- Priority Date
- 2020-10-30
- Filing Date
- 2021-10-19
- Publication Date
- 2024-12-04
- Estimated Expiration
- 2041-10-19
AI Technical Summary
High modulation frequency in indirect time-of-flight depth mapping systems results in high direct and capacitive currents, leading to significant electromagnetic interference and voltage disruptions, which degrade measurement accuracy and synchronization.
Spatially allocating phase shifts on the demodulation signal across a matrix of photosensitive pixels, distributing phase shifts at different values within groups of pixels to reduce current peaks and electromagnetic interference, and compensating for these shifts in phase shift calculations to maintain measurement accuracy.
This approach reduces the intensity and frequency of electromagnetic interference, allowing for higher resolution and modulation frequency in depth mapping systems while minimizing disruptions to the measurement process.
Description
[0001] Methods of implementation and embodiment relate to integrated circuits, in particular integrated sensors for acquiring depth mapping by indirect time of flight, usually designated by the acronym "iToF" from the English term "indirect Time of Flight".
[0002] Traditionally, time-of-flight measurement systems measure the time between the instant of emission of an optical signal, i.e. a light signal, and the instant of reception of this signal after reflection on elements present in the detection field, i.e. the field of vision of the system in the illuminated area. The distance separating the elements of the detection field and the system is reconstructed from the measured time, proportionally to the speed of light.
[0003] Indirect time-of-flight (iToF) systems transmit an optical signal modulated at a modulation frequency, and measure the phase shift of the modulation of the received signal relative to the transmitted signal. The phase shift measurement can be obtained from the amount of charge photogenerated by pixels of a receiver during integration periods controlled by a demodulation signal synchronous with the modulation of the optical signal.
[0004] Two out-of-phase samplings (i.e., two successive integrations, respectively controlled by the in-phase demodulation signal and by the out-of-phase demodulation signal) are sufficient to reconstruct the phase shift.
[0005] That being said, techniques using more than two samplings, such as for example the technique known as “4-bin sampling” using four samplings with differential components in phase quadrature of the demodulation signal (i.e. differential components at 0° and 180° and differential components in phase quadrature at 90° and 270°), are more commonly implemented because they have advantages, in particular due to the known and controlled mechanisms for demodulating signals in phase quadrature (“I / Q” in the field of radiofrequency communications).
[0006] Publication WO 2020 / 089062 A1 describes an electronic device having an array of pixels, and pixel modulation driving circuits, each pixel modulation driving circuit being configured to drive a sub-array of the pixels; in particular in the context of the 4-bin sampling technique.
[0007] Publication WO 2020 / 089057 A1 describes an electronic device having an array of lighting units and multiple control circuits, each control circuit being adapted to drive a sub-array of the lighting units.
[0008] Increasing the modulation frequency improves the accuracy of the measured distance, but also increases the ambiguity of the measurement (i.e., the periodicity of different distances resulting in the same phase shift), which reduces the measurement range. However, measurement techniques at several modulation frequencies can limit the problems related to ambiguity.
[0009] It is desirable to propose indirect time-of-flight depth mapping acquisition systems with high resolution, for example of the order of megapixels, and with a high modulation frequency, for example greater than 100MHz (megahertz) or even greater than 200MHz.
[0010] However, increasing the modulation frequency and resolution of conventional systems produces very high direct currents, for example close to 2A (amperes) for a 1Megapixel sensor modulated at 200MHz, during the integration phases and also capacitive current peaks close to 25A over a hundred picoseconds.
[0011] These current peaks generate, on the one hand, very strong electromagnetic interference (EMI) and, on the other hand, cause drops in the supply voltage and increases in the ground voltage, which can in particular disrupt the switching of logic elements and degrade the synchronization of the time-of-flight measurement.
[0012] Therefore, there is a need to propose solutions to address the problems caused by the above-mentioned current spikes in high-resolution, high-modulation-frequency indirect time-of-flight depth mapping acquisition systems.
[0013] According to embodiments and modes of implementation, it is proposed to spatially allocate phase shifts on the demodulation signal in a matrix of photosensitive pixels, in order to spread electromagnetic interference. The spatial allocation of the phase shifts can be defined on a pre-established pixel pattern, which can be one or more columns, one or more rows, or a rectangular sub-array of pixels.
[0014] The invention is defined by the independent claims of the appended set of claims.
[0015] Modes of implementation and embodiment of the invention are defined by the dependent claims of the appended set of claims.
[0016] According to one aspect, there is provided a method for acquiring a depth map by indirect time of flight in an array of photosensitive pixels segmented into groups of pixels, comprising at least one capture during which the pixels of the array are controlled by a demodulation signal, the method comprising an introduction of phase shifts in the demodulation signal at different values and distributed in each group of pixels.
[0017] For example, if the acquisition comprises four captures during which the demodulation signal used is provided in phase opposition or in phase quadrature, the introduction of phase shifts at different values and distributed in each group of pixels is carried out on the demodulation signals having the respective phase shifts in opposition or in quadrature at each capture.
[0018] Thus, due to the distribution of phase shifts on the demodulation signal in the groups of pixels of the network, the pixels of the network are not all controlled at the same times by a single demodulation signal during a capture, but the times at which the pixels are controlled are distributed in time.
[0019] Therefore, the current peaks consumed in the network are of lower intensity and are distributed over time, and the amplitude of the resulting electromagnetic interference is reduced proportionally to the number of different phase shifts. In addition, the frequency of the electromagnetic interference is shifted proportionally to the number of different phase shifts.
[0020] According to one implementation mode, the values of the phase shifts are discretely distributed over a period of the demodulation signal.
[0021] For example, "discretely distributed" means a distribution in a quantity deliberately limited to a number much lower than the maximum quantity of different shifts that can be introduced in theory.
[0022] According to one embodiment, the array is arranged in columns and rows of pixels, and the number of discrete phase shift values is between substantially one hundredth and substantially one tenth of the number of columns or rows. By "substantially" is meant, for example, "rounded down or up to the nearest integer."
[0023] Indeed, providing a discrete number of phase shifts presents implementation advantages, notably with simpler and less expensive means, but also in a more easily controllable manner from the point of view of the calibration of the phase shifts introduced, that is to say the precision of the generation of the phase shifts.
[0024] Furthermore, the phase shift values may or may not be uniformly distributed over a period of the demodulation signal. However, a uniform distribution of phase shifts over a period of the demodulation signal allows for optimal minimization of electromagnetic interference.
[0025] And, distributing the phase shift values relative to a period of the demodulation signal makes it possible to control the offset frequency of the resulting electromagnetic interference, and to achieve a reproducible spectral spread regardless of the modulation frequency, unlike, for example, arbitrary phase shifts which would lead to an uncontrolled spectral spread. This is all the more advantageous in the context of an acquisition at several modulation frequencies.
[0026] According to one embodiment, the number of different values of the phase shifts is chosen so that the product of the frequency of the demodulation signal with said number is located outside a bandwidth of interest.
[0027] Indeed, electromagnetic interference can be particularly problematic for certain frequencies, for example in a bandwidth of interest containing the frequencies at which systems neighboring the system implementing the acquisition of depth mapping operate. Also, certain materials of the system implementing the acquisition of depth mapping can, by nature, attenuate frequencies beyond a given bandwidth, constituting the bandwidth of interest in the context of this mode of implementation.
[0028] According to one embodiment, the array is arranged in columns and rows of pixels and the groups of pixels are segmented in a periodic pattern across the columns and / or rows.
[0029] Choosing a periodic pattern in the pixel array plane allows current consumption to be distributed in space, limiting the appearance of constructive phenomena in the generation of electromagnetic interference, and standardizing the system's performance.
[0030] According to one embodiment, the groups of pixels are segmented according to a periodic pattern of one or more columns such that each group comprises several spatially non-consecutive columns or several spatially non-consecutive blocks of spatially consecutive columns.
[0031] This mode of implementation is advantageous in terms of signal distribution in the case where the architecture of the pixel network provides for distribution of the demodulation signal by columns of pixels.
[0032] According to one embodiment, the method further comprises a calculation of a time-of-flight phase shift, for each pixel, between the demodulation signal and a light signal received during said at least one capture, and the calculation of the time-of-flight phase shift comprises a compensation for the phase shift introduced on the demodulation signal for each group of pixels.
[0033] According to one embodiment, said compensation comprises, for each group of pixels, a modulo 360° addition of the value of the phase shift introduced on the corresponding demodulation signal, to the calculated time-of-flight phase shift.
[0034] In other words, the compensation is performed after calculating the phase shift seen during the integration of the optical signal by the pixels, during the reconstruction of the distance, typically by digital means that can be easily parameterized in this regard.
[0035] According to one embodiment, the calculation of the time-of-flight phase shift comprises a trigonometric operation performed on arguments resulting from at least two captures during which the pixel array is controlled by a respective demodulation signal, and in which said compensation comprises, for each group of pixels, a rotation of the arguments of the trigonometric operation by an angle equivalent to the value of the phase shift introduced on the corresponding demodulation signal.
[0036] In other words, the compensation is performed during the trigonometric calculation of the phase shift seen during the integration of the optical signal by the pixels, which makes it possible to directly provide the time-of-flight phase shift taking into account the compensation, to reconstruct the distance.
[0037] According to one embodiment, the method further comprises an emission of a light signal modulated by a modulation signal, the demodulation signal being synchronous with the modulation signal.
[0038] According to another aspect, there is provided an integrated circuit suitable for indirect time-of-flight depth mapping, comprising an array of photosensitive pixels segmented into groups of pixels, control means configured to control the pixels of the array with a demodulation signal during at least one capture of a depth mapping acquisition, the control means being configured to introduce phase shifts in the demodulation signal at different values and distributed in each group of pixels.
[0039] According to one embodiment, the control means are configured to introduce phase shifts at values discretely distributed over a period of the demodulation signal.
[0040] According to one embodiment, the array is arranged in columns and rows of pixels, and the control means are configured to introduce a number of discrete phase shift values between substantially one hundredth and substantially one tenth of the number of columns or rows.
[0041] According to one embodiment, the control means are configured to introduce a number of different values of the phase shifts chosen so that the product of the frequency of the demodulation signal with said number is located outside a bandwidth of interest.
[0042] According to one embodiment, the array is arranged in columns and rows of pixels and the groups of pixels are segmented in a periodic pattern across the columns and / or across the rows.
[0043] According to one embodiment, the groups of pixels are segmented according to a periodic pattern of one or more columns such that each group comprises several spatially non-consecutive columns or several spatially non-consecutive blocks of spatially consecutive columns.
[0044] According to one embodiment, the integrated circuit further comprises calculation means configured to calculate a time-of-flight phase shift, for each pixel, between the demodulation signal and a light signal received during said at least one capture, and the calculation means are configured to compensate for the phase shift introduced by the control means on the demodulation signal for each group of pixels.
[0045] According to one embodiment, the calculation means are configured to compensate for the phase shift introduced into the demodulation signal, for each group of pixels, by adding modulo 360° the value of the corresponding phase shift, to the calculated time-of-flight phase shift.
[0046] According to one embodiment, the control means are configured to control the pixels of the array with respective demodulation signals during at least two captures of a depth mapping acquisition, the calculation means are configured to calculate the time-of-flight phase shift by performing a trigonometric operation on arguments resulting from said at least two captures, and the calculation means are configured to compensate for the phase shift introduced on the demodulation signals, for each group of pixels, by pivoting by an angle equivalent to the value of the corresponding phase shift, the arguments of the trigonometric operation.
[0047] According to one embodiment, the integrated circuit further comprises an emission means configured to emit a light signal modulated by a modulation signal, and the control means are configured to generate the modulation signal and the demodulation signal synchronous with the modulation signal.
[0048] Other advantages and characteristics of the invention will appear on examining the detailed description of embodiments and implementations, which are in no way limiting, and the appended drawings, in which: [ Fig 1 ] ; [ Fig 2 ] ; [ Fig 3 ] ; [ Fig 4 ] ; [ Fig 5 ] illustrate embodiments and implementations of the invention.
[0049] There figure 1 illustrates an example of an integrated CI sensor of the indirect time of flight type “iToF” (acronym for the common English terms “indirect Time of Flight”), intended to measure a depth map (usually “depth map” in English).
[0050] The integrated sensor IC comprises an EM optical transmitter for transmitting a modulated optical signal at a modulation frequency. The EM optical transmitter comprises, for example, an infrared laser diode, and is modulated by a periodic MOD modulation signal, typically a step signal, at a modulation frequency, for example, greater than 100 MHz.
[0051] CMD control means are configured to control illumination phases by the optical transmitter EM and in particular to generate the modulation signal MOD.
[0052] The CMD control means are also configured to control an integration phase, also called capture, of an optical signal resulting from the reflection of the modulated optical signal emitted during illumination, by an array of photosensitive pixels RES_PX.
[0053] Each capture is performed simultaneously with illumination, and the pixel integration times are controlled by a DEMOD demodulation signal. The DEMOD demodulation signal is synchronous with the MOD modulation signal, and makes it possible to measure the phase shift of the received optical signal compared to the emitted optical signal in order to deduce the distance separating the object of the reflection and the CI sensor.
[0054] The RES_PX photosensitive pixel array comprises pixels that are photosensitive to the wavelength of the emitted optical signal, and can have a density of the order of 1MP (MegaPixel), for example 0.5MP. The density of the RES_PX array is also called "definition".
[0055] The RES_PX pixel array is arranged into columns of pixels accessible in particular via a DECY column decoder, and into rows of pixels accessible via a DECX row decoder.
[0056] The CMD control means are configured to control the pixels of the RES_PX network with a DEMOD demodulation signal during captures, for example two or four captures, as part of the acquisition of a depth map.
[0057] Furthermore, calculation means CAL are configured to calculate the phase shift Φ, also called time-of-flight phase shift, between the modulation of the emitted optical signal and the modulation of the received optical signal during said at least one capture, from the quantities of charges photogenerated by the pixels of the RES_PX network controlled by the demodulation signal DEMOD.
[0058] The quantities of charges photogenerated by the pixels can typically be supplied to the CAL calculation means by voltage signals from an analog-to-digital converter incorporated in the DECY column decoder.
[0059] The DECY column decoder can also be provided for typical processing such as dark current subtraction, in addition to its decoding function in cooperation with the DECX row decoder and following the RES_PX pixel array architecture.
[0060] Furthermore, the pixel array RES_PX is segmented into pixel blocks, e.g. fifty-six blocks B1, B2, ..., B56, each comprising several columns, e.g. twelve columns Col1-Col12 ( figure 2 ).
[0061] Segmentation into pixel blocks could be done on rows of the RES_PX array or on both rows and columns, i.e. on rectangular portions of the RES_PX pixel array.
[0062] It will be noted that the segmentation into blocks of pixels is provided within the framework of a distribution of the demodulation signal and can be purely virtual, that is to say that the pixel network RES_PX does not necessarily include a structure specific to delimiting the blocks of the segmentation, unlike for example the columns and the rows which correspond in particular to electrically conductive tracks contacting each pixel of the network RES_PX.
[0063] The CMD control means are in fact configured to introduce phase shifts Δ1-Δ12 ( figure 2 ) in the DEMOD demodulation signal at different values and distributed in respective pixel groups.
[0064] From a hardware point of view, the CMD control means can be implemented by logic circuits and include a state machine, in order to implement the control and timing functions of the CI sensor, and can further include signal generation means configured to generate in particular the modulation MOD and demodulation DEMOD signals, and introduce the phase shifts Δ1-Δ12.
[0065] In this regard, reference is made to the figure 2 .
[0066] There figure 2 represents any block Bk among the pixel blocks B1-B56 of the example of the figure 1 . In this example, block Bk has twelve columns Col1-Col12 of pixels.
[0067] In this example, each photosensitive pixel is of the “2-tap” type, i.e. capable of photogenerating charges, at the same time and distinctly during two consecutive half-periods of the DEMOD demodulation signal.
[0068] The operation of the 2-tap pixels is schematized by two inputs TAP1 and TAP2 receiving the demodulation signal DEMOD in phase opposition, for example 0° on the TAP1 input and 180° on the TAP2 input in the first column Col1.
[0069] That being said, it will be necessary to make the distinction (particularly in relation to the figure 4 ) between the charge photogeneration of the 2-tap mechanism and a second capture Capt2 ( figure 4 ) controlled by the DEMOD demodulation signal in phase opposition. Indeed, the photogeneration of charges of the 2-tap mechanism can correspond to information acquired in a “differential” manner when controlling a single DEMOD demodulation signal, while the second capture Capt2 ( figure 4 ) is controlled with a “reference” demodulation signal DEMOD initially planned in phase opposition with the modulation signal MOD.
[0070] The CMD control means are configured to generate the demodulation signal DEMOD, called “reference”, perfectly synchronous with the modulation signal MOD controlling the emission of the optical signal. By “synchronous”, we mean for example that the phase of the reference demodulation signal DEMOD is locked to the phase of the modulation signal MOD (possibly in phase opposition or in phase quadrature during successive captures of an acquisition, see below in relation to the figure 4 ).
[0071] The control means are configured to further introduce phase shifts Δ1-Δ12 into the reference demodulation signal DEMOD, resulting in shifted demodulation signals DEMOD+Δi, with i=[1:12], for example by means of a phase-locked loop PLL or a phase generator PGEN.
[0072] Each column Coli, i=[1:12], of block Bk receives a respective shifted demodulation signal DEMOD+Δi, i.e. with phase shifts Δ1-Δ12 of different values and distributed in each column of pixels Col1-Col12.
[0073] In this advantageous example, the phase shifts are introduced into the reference demodulation signal DEMOD at values discretely and uniformly distributed over a period of the demodulation signal, i.e., for example, twelve shifts from 0° to 330° with a step of 30°.
[0074] This corresponds, on the TAP2 input in phase opposition, to six shifts from 180° to 330° with the 30° step and six shifts from 0° to 150° with the 30° step.
[0075] Other discrete amounts of phase shift values may be provided, for example, sixty-four (64) shifts with a step size of 5.625°.
[0076] From a more general point of view, we can predict the quantity of discrete phase shift values between one hundredth of the total number of columns in the RES_PX network and one tenth of the total number of columns in the RES_PX network, for example rounded down or up to the nearest integer. In the example shown in the figures 1 And 2 , this corresponds to a quantity between 6 and 68 different discrete values.
[0077] If the pixel groups in which the different phase shift values are distributed are formed relative to the rows of the RES_PX array, then the quantity of discrete shift values can be defined in the same way, relative to the total number of rows of the RES_PX array.
[0078] It is indeed advantageous to provide a "discrete" quantity of shifts, in order to limit the complexity of generating phase shifts. For example, "discrete quantity" means a quantity deliberately limited to a number much lower than the maximum quantity of different shifts that can be introduced in theory, i.e. for example the total number of columns in the RES_PX pixel array, or even in absolute terms the total number of pixels in the RES_PX array.
[0079] In fact, the CMD control means generating the phase shifts can thus be easily produced, in a compact manner, and in addition be precisely calibrated to the desired phase shift values and according to the desired step.
[0080] In fact, this distribution of phase shifts, by column within a Bk block, is applied in all blocks B1-B56 of the RES_PX pixel array of the CI sensor during capture.
[0081] In other words, the pixels of each individual column Coli, i=[1:12], in all blocks Bk, k=[1:56], are driven during capture with the same shifted demodulation signal DEMOD+Δi.
[0082] Thus, the phase shifts Δi are introduced into the demodulation signal DEMOD at different values and distributed in respective pixel groups, a pixel group corresponding in this example to several spatially non-consecutive columns (one column Coli, in each block Bk). In other words, the pixel groups are segmented according to a one-column pattern, with a periodicity of one block.
[0083] That being said, the pixel groups can be segmented in a periodic pattern of several columns, so that each group has several spatially non-consecutive sets of spatially consecutive columns, such as in the example illustrated by figure 4 .
[0084] Thus, during a capture, the distribution of phase shifts on the demodulation signal in the different groups of pixels of the network results in commands of the different groups of pixels of the network which are triggered at times distributed in time.
[0085] Alternatively, the pixel groups may be segmented according to a pattern having no spatial periodicity, for example according to a pseudo-random, "mixed" distribution in the RES_PX pixel array. The distribution of phase shifts on the demodulation signal in such a pseudo-random distribution of the different pixel groups of the RES_PX array results in the same effects of distribution in time of the command triggers.
[0086] Indeed, the DEMOD demodulation signal controls transistor switching and charge transfers in the pixels of the network. The currents flowing in the RES_PX pixel network resulting from these acquisition mechanisms, in particular capacitive currents, can be very large and generate electromagnetic interference.
[0087] However, the distribution of switching times over time makes it possible to reduce electromagnetic interference.
[0088] In this regard, reference is made to the figure 3 .
[0089] There figure 3 schematically illustrates on a first graph G1 the generation of current peaks Ipk as a function of time T in the RES_PX network during a capture, and on a second graph G2 the EMI amplitude and the frequency F of the electromagnetic interference which result from the current peaks.
[0090] In both graphs G1, G2, the dashed curves illustrate the classic case in which all the pixels of the RES_PX network are controlled at the same time by the reference demodulation signal DEMOD.
[0091] The continuous line curves illustrate the effect of the time distribution of the control instants, obtained by the introduction of phase shifts in the demodulation signal DEMOD at different values Δ1-Δ12 and distributed in each group of pixels, as previously described in relation to the figures 1 And 2 .
[0092] In the classical case, high intensity current peaks Ip, for example up to 25A (ampere), are generated periodically at the period 1 / fmod of the reference demodulation signal DEMOD, with fmod the frequency of the reference demodulation signal, for example greater than 100MHz.
[0093] As a result, electromagnetic interference is generated at frequency fmod with a large amplitude A.
[0094] When the pixels are controlled by the shifted demodulation signals DEMOD+Δi, with N different phase shifts and distributed in N groups of pixels of the RES_PX network, there are N times fewer switchings at the same instant, but N times more different instants of switching of groups of pixels.
[0095] Therefore, the intensity of the current peaks Ip / N is approximately equal to the intensity of the classical case Ip divided by N, and the period of generation of the current peaks is also divided by N, 1 / (N*fmod) compared to the classical case.
[0096] As a result, electromagnetic interference is generated at a frequency N*fmod shifted by a factor of N, and at a lower amplitude A / D.
[0097] Thus, the embodiments and implementation methods described in relation to the figures 1 And 2 result in peak currents consumed in the network of lower intensities and distributed discretely in time. The amplitude of the electromagnetic interference generated by the peak currents is reduced proportionally to the number of different phase shifts, and the frequency of the electromagnetic interference is shifted proportionally to the number of different phase shifts.
[0098] On the one hand, the number N of different values of the phase shifts Δi introduced into the demodulation signal DEMOD can be chosen so that the A / D amplitude of the electromagnetic interference is sufficiently low so as not to harm the operation of the CI sensor or its vicinity.
[0099] On the other hand, we can also choose the number N of different values of the phase shifts Δi introduced into the demodulation signal DEMOD so that the frequency of the electromagnetic interference N*fmod is located outside a bandwidth of interest.
[0100] The bandwidth of interest may, for example, correspond to an operating frequency band of the IC sensor, so that electromagnetic interference with frequencies outside this range does not disrupt the operation of the IC sensor. Also, structural elements of the IC sensor, such as Faraday cage shields in the encapsulation packages, can absorb and attenuate electromagnetic waves located outside the bandwidth of interest. In the second case, electromagnetic interference whose amplitude is already divided by N can thus be attenuated additionally.
[0101] It should be noted that the representations of graphs G1 and G2 correspond to an introduction of the phase shift values uniformly distributed over a period of the demodulation signal. This results in particular in a frequency spectrum of electromagnetic interference positioned only on the frequency equal to the product N*fmod mentioned above.
[0102] However, a non-uniform distribution of phase shifts over a period of the demodulation signal would cause a wider spectral distribution of electromagnetic interference.
[0103] In this case, we can jointly choose the non-uniform distribution of phase shifts and the number N so that the frequency spectrum of electromagnetic interference, or a major part of the spectrum, is located outside the bandwidth of interest.
[0104] Thus, the embodiments and implementation methods described in relation to the figures 1 And 2 allow the removal of constraints relating to electromagnetic interference, which makes possible increases in the resolution of the pixel array and the frequency of the modulation signal, in indirect time-of-flight depth mapping acquisition systems.
[0105] It is recalled that the indirect time-of-flight distance measurement is based on a time-of-flight phase shift of the received optical signal, evaluated from the demodulation signal, and we now refer to the figures 4 And 5 illustrating embodiments and implementations for compensating for the phase shift introduced into the demodulation signal for each pixel group in the evaluation of the time-of-flight phase shift.
[0106] There figure 4 illustrates on the one hand another example of segmentation of the network of photosensitive pixels into groups of pixels, and on the other hand a method of compensating for the phase shifts Δi introduced on the demodulation signal DEMOD.
[0107] In this example, the pixel array RES_PX is first segmented into column blocks Bk as described previously in relation to the figure 1 , each block Bk having twelve columns Col1-Col12.
[0108] In this example, a pixel group has two consecutive columns Col1 / 2, Col3 / 4, ..., Col11 / 12, in each block Bk.
[0109] Therefore, pixel groups are segmented in a periodic pattern of two columns, so that each group has several spatially non-consecutive sets of two spatially consecutive columns.
[0110] On the other hand, the figure 4 also illustrates a procedure for acquiring depth mapping, comprising four successive captures Capt1, Capt2, Capt3, Capt4, using a 4-bin sampling technique whose principle is known to those skilled in the art.
[0111] In summary, the 4-bin sampling technique uses four captures, i.e., four integrations with respective illuminations, driven with quadratic phase-shifted demodulation signal components.
[0112] In other words, in each capture Capt1, Capt2, Capt3, Capt4, the reference demodulation signal is respectively set to phases of 0°, 180°, 90°, 270° of the modulation signal MOD used for the emission of the optical signal.
[0113] The quantity of charges photogenerated by each pixel during a capture, usually called "bin", provides information corresponding to the superposition of the phase shift of the received optical signal with the phase of the demodulation signal.
[0114] For example, during the first capture Capt1 the pixels provide information about the in-phase demodulation signal Bin_0; during the second capture Capt2 the pixels provide information about the out-of-phase demodulation signal Bin_180; during the third capture Capt3 the pixels provide information about the quadrature-phase demodulation signal Bin_90; during the fourth capture Capt4 the pixels provide information about the out-of-phase quadrature-phase demodulation signal Bin_270.
[0115] As an aside, the "2-tap" operation of the pixels allows each Bin value to be collected twice in the two captures in respective phase opposition (for example, the Bin_0 information is collected by the TAP1 of the first capture Capt1 and by the TAP2 of the second capture Capt2). This is particularly advantageous in terms of ambient noise (usually "offset" in English) and measurement concordance (usually "matching" in English).
[0116] The reconstruction of the phase shift Φ, also called "time-of-flight phase shift", is done by CAL calculation means and includes, for each pixel, a TRIGO trigonometric operation of the type "arctan(Q / I)" with Q an argument representing the quadrature phase shift obtained by trigonometric relation on the quantities Bin_90 and Bin_270, and I an argument representing the in-phase phase shift obtained by trigonometric relation on the quantities Bin_0 and Bin_180.
[0117] Now, the phase shift Δi respectively introduced on the modulation signal DEMOD+Δi of each group of pixels corresponds to a rotation of an angle known in the arguments Q and I.
[0118] Therefore, to compensate for the phase shift Δi introduced on the demodulation signal DEMOD+Δi, it is possible to adapt the trigonometric calculation, for each group of pixels, by pivoting the arguments Q and I by an angle equivalent to the value of the phase shift Δi respectively introduced.
[0119] The distance can then be directly calculated knowing the phase shift Φ compensated for each group of pixels.
[0120] There figure 5 illustrates another method of compensating for phase shifts Δi introduced into the DEMOD demodulation signal.
[0121] In this method, the calculation means CAL are configured to obtain the value of the phase shift Φ of each pixel by a classic trigonometric calculation, that is to say independently of the group to which the pixel belongs and without taking into account the phase shift Δi introduced on the demodulation signal DEMOD+Δi relating to this group.
[0122] The CAL calculation means are configured to compensate for each phase shift Δi introduced on the demodulation signal, on the calculated phase shift value Φ. As the calculation of the phase shift Φ was made relative to a shifted signal DEMOD+Δi, the CAL calculation means add the value of the shift Δi corresponding to the phase shift value Φ.
[0123] If the calculated phase shift Φ is between 0° and 360°-Δi, then the actual phase shift Φout is Φ+Δi; and if the calculated phase shift Φ is between 360°-Δi and 360°, then the actual phase shift Φout is Φ-360°+Δi. This amounts to performing a modulo 360° addition of the value of the phase shift Δi to the calculated time-of-flight phase shift Φ.
[0124] Furthermore, the invention is not limited to these embodiments and implementations but embraces all variants thereof, for example, the introduction of phase shifts in the demodulation signal at different values and distributed in groups of pixels can be adapted to other mechanisms than the mechanisms described, in particular the 4-bin sampling technique of the acquisition or the 2-tap mechanism of the pixels, which have only been given as advantageous and in no way limiting examples.
Claims
1. A method for acquiring a depth map by indirect time-of-flight in a network of photosensitive pixels (RES_PX) segmented into groups of pixels, comprising at least one capture (Capt1-Capt4) during which the pixels of the network are controlled by a demodulation signal (DEMOD), characterised in that the method comprises introducing phase shifts into the demodulation signal at different values (Δ1-Δ12) distributed in each group of pixels (Coll-Col12, Bk), and in that the method further comprises calculating, for each pixel, a time-of-flight phase difference (φ) between the demodulation signal (DEMOD) and a light signal received during said at least one capture (Capt1-Capt4), the calculation of the time-of-flight phase difference (φ) comprising a compensation for the phase shift (Δi) introduced onto the demodulation signal for each group of pixels.
2. The method according to claim 1, wherein the values of the phase shifts (Δ1-Δ12) are discretely distributed over a period of the demodulation signal (DEMOD).
3. The method according to claim 2, wherein the network (RES_PX) is arranged into columns and into rows of pixels, and the number of discrete values of the phase shifts (Δ1-Δ12) is comprised between substantially one hundredth and substantially one tenth of a number of columns or of rows.
4. The method according to one of the preceding claims, wherein the number (N) of different values of the phase shifts (Δ1-Δ12) is selected so that the product of a frequency of the demodulation signal and said number (N*fmod) is outside a bandwidth of interest.
5. The method according to one of the preceding claims, wherein the network (RES_PX) is arranged into columns and into rows of pixels, and the groups of pixels are segmented according to a periodic pattern over the columns and / or over the rows.
6. The method according to claim 5, wherein the groups of pixels are segmented according to the periodic pattern of one or more column(s) so that each group includes several spatially non-consecutive columns or several spatially non-consecutive sets of spatially consecutive columns.
7. The method according to one of claims 1 to 6, wherein said compensation comprises, for each group of pixels, a modulo 360° addition (ADD[360°]) of the value of the phase shift (Δi) introduced onto the corresponding demodulation signal, to the calculated time-of-flight phase difference (φ).
8. The method according to one of claims 1 to 6, wherein the calculation of the time-of-flight phase difference (φ) comprises a trigonometric operation (TRIGO) performed on arguments (I, Q) resulting from at least two captures (Capt1-Capt4) during which the network of pixels (RES_PX) is controlled by a respective demodulation signal, and wherein said compensation comprises, for each group of pixels, a rotation (ROT[Δi]) of the arguments of the trigonometric operation by an angle equivalent to the value of the phase shift (Δi) introduced onto the corresponding demodulation signal.
9. The method according to one of the preceding claims, further comprising emitting a light signal modulated by a modulation signal (MOD), the demodulation signal (DEMOD) being synchronous with the modulation signal (MOD).
10. An indirect time-of-flight type sensor (CI) adapted for depth mapping, including a network of photosensitive pixels (RES_PX) segmented into groups of pixels, control means (CMD) configured to control the pixels of the network with a demodulation signal (DEMOD) during at least one capture (Capt1-Capt4) of an acquisition of a depth map, characterised in that the control means (CMD) are configured to introduce phase shifts into the demodulation signal at different values (Δ1-Δ12) distributed in each group of pixels, and in that the sensor further includes calculation means (CAL) configured to calculate a time-of-flight phase difference (φ), for each pixel, between the demodulation signal (DEMOD) and a light signal received during said at least one capture (Capt1-Capt4), the calculation means (CAL) being configured to compensate for the phase shift (Δi) introduced by the control means (CMD) onto the demodulation signal (DEMOD) for each pixel group.
11. The sensor according to claim 10, wherein the control means (CMD) is configured to introduce phase shifts at values (Δ1-Δ12) discretely distributed over a period of the demodulation signal (DEMOD).
12. The sensor according to claim 11, wherein the network (RES_PX) is arranged into columns and into rows of pixels, and the control means (CMD) are configured to introduce a number (N) of discrete values of phase shifts (Δi) comprised between one hundredth and one tenth of the number of columns or of rows.
13. The sensor according to one of claims 10 to 12, wherein the control means (CMD) are configured to introduce a number (N) of different values of the phase shifts (Δi) selected so that a product of the frequency of the demodulation signal and said number (N*fmod) is outside a bandwidth of interest.
14. The sensor according to one of claims 10 to 13, wherein the network (RES_PX) is arranged into columns and into rows of pixels, and the groups of pixels are segmented according to a periodic pattern over the columns and / or over the rows.
15. The sensor according to claim 14, wherein the groups of pixels are segmented according to a periodic pattern of one or more column(s) so that each group includes several spatially non-consecutive columns or several spatially non-consecutive sets of spatially consecutive columns.
16. The sensor according to one of claims 10 to 15, wherein the calculation means (CAL) are configured to compensate for the phase shift (Δi) introduced onto the demodulation signal (DEMOD), for each group of pixels, by modulo 360° addition of the value of the corresponding phase shift (ADD[360°]), to the calculated time-of-flight phase difference.
17. The sensor according to one of claims 10 to 15, wherein the control means (CMD) are configured to control the pixels of the network (RES_PX) with respective demodulation signals during at least two captures (Capt1-Capt4) of a depth map acquisition, the calculation means (CAL) are configured to calculate the time-of-flight phase difference (φ) by performing a trigonometric operation (TRIGO) on arguments (I, Q) resulting from said at least two captures, and wherein the calculation means (CAL) are configured to compensate for the phase shift (Δi) introduced onto the demodulation signals (DEMOD), for each group of pixels, by pivoting by an angle equivalent to the value of the corresponding phase shift (ROT[Δi]), the arguments of the trigonometric operation.
18. The sensor according to one of claims 10 to 17, further comprising an emission means (EM) configured to emit a light signal modulated by a modulation signal (MOD), the control means (CMD) being configured to generate the modulation signal (MOD), the demodulation signal (DEMOD) synchronous with the modulation signal (MOD).
