CN115996325B - Hilbert curve-based SPAD array and imaging method - Google Patents

Abstract

The invention discloses a SPAD array based on a Hilbert curve and an imaging method, and belongs to the technical field of semiconductor devices and the technical field of image communication. The invention comprises a pixel array formed by a plurality of pixel units, a Hilbert linear array, a head reading module and a tail reading module; the Hilbert points refer to points on the Hilbert linear array which are arranged in the form of Hilbert curves; the output of each pixel is connected with a Hilbert point; the head reading module and the tail reading module are respectively connected with the starting point and the end point of the Hilbert linear array; the hilbert linear array is responsible for pulse signal transmission. In the invention, imaging can be performed by only measuring the head-to-tail signals of the Hilbert linear array. Compared with the traditional SPAD array with the counting circuit, the pixel array is greatly reduced, the pixel filling factor is higher, and the ultra-high-speed reading can be realized; the invention greatly reduces the area, and is especially suitable for low-cost, low-power consumption, low-illumination and ultra-high-speed systems.

Description

Hilbert curve-based SPAD array and imaging method

Technical Field

The invention relates to a Hilbert curve-based SPAD array and an imaging method, and belongs to the technical field of semiconductor devices and the technical field of image communication.

Background

According to quantum theory, photons are the smallest carrier of light energy, so the theoretical limit of the sensitivity of a photodetector is the ability to detect a single photon. SPAD (single photon avalanchediode ) is a photodiode operating under extremely high reverse bias, when a single photon is incident and generates a carrier, the photon-generated carrier is continuously accelerated and collided under the action of the reverse bias to generate a new electron-hole pair, so that an avalanche current signal is generated in extremely short time (generally in ps magnitude), and therefore, the SPAD has single photon detection capability and can accurately measure the arrival time of the photon through the change of an electric signal.

The characteristic advantage of a single SPAD can be integrated with the array. Single photon avalanche photodiode (SPAD) arrays are favored for applications such as 3D imaging, fluorescence lifetime imaging, lidar, etc., and future research scenarios are wide, but often suffer from low fill factor (referring to the ratio of the area of the photosensitive area to the area of the picture element). The value of the SPAD fill factor (the fill factor is the ratio of the photosensitive area to the pixel size) is usually less than 5%, because the counting circuit is actively large, if a part of performance is sacrificed, such as deleting the counting circuit, the SPAD array fill factor can be as high as 60%, and if the unit is deleted, the common practice in the industry is off-chip counting, the gray information is reflected by obtaining the counter value of the unit exposure time, the practice has a great limitation, the off-chip counting can only be read one by one, the number of pixels in the array is increased by a factor of the off-chip counting time, the chip area is reduced, the cost is lower, and the ultra-high speed application is limited greatly. At present, ultra-high speed and high filling factors are not available in the SPAD array.

The SPAD array integrated with a plurality of pixels has parallel single photon information acquisition capability, and is superior to the traditional single-point scanning architecture in detection efficiency. In addition, the photon counting circuit and the time measuring circuit can be integrated into the SPAD array, so that the functionality and the expandability of the SPAD array are obviously improved. However, this method has a great limitation, the area of the counter is very large, and the circuit used in the design is more complex, so eliminating the counter becomes a great difficulty in the prior art.

Disclosure of Invention

In order to overcome at least one of the above mentioned problems, such as the ultra-high speed and the unavailability of high filling factor in SPAD arrays, and eliminate the problems of counters, the invention provides a SPAD array based on hilbert curve and an imaging method, which can complete the ultra-high speed application while the pixels have high filling factor. Compared with the traditional SPAD array with the counting circuit, the invention only needs the first and the last reading circuits, the pixel array under the same pixel is greatly reduced, and the pixel filling factor is higher; compared with an off-chip counting SPAD array, the array can realize ultra-high-speed reading, and simultaneously collect the gray scale of all pixels of the whole array.

The Hilbert Curve (Hilbert Curve) adopted by the invention is a filling Curve. The Hilbert curve, depending on the characteristics of its own space-filling curve, can linearly traverse each discrete unit in two or higher dimensions and traverse each discrete unit only once, and linearly sequence and encode each discrete unit as a unique identification of that unit. The space filling curve can map data which has no good sequence in a high-dimensional space to a one-dimensional space, and through the coding mode, objects which are adjacent in space can be stored in a block, so that IO time can be reduced, and the data processing efficiency in the memory is improved.

In order to achieve the above purpose, the technical scheme adopted by the invention is as follows:

the invention discloses a hilbert curve-based SPAD array, which comprises a pixel array (SPAD array) formed by a plurality of pixel units, a hilbert linear array, a head reading module and a tail reading module; the Hilbert linear array is a linear array which is arranged in a Hilbert curve mode, and Hilbert points are points on the Hilbert linear array; the output of each pixel is connected with a Hilbert point; the first reading module is connected with the starting point of the Hilbert linear array, and the tail reading module is connected with the end point of the Hilbert linear array; the hilbert linear array is responsible for pulse signal transmission.

The pixel unit comprises a Single Photon Avalanche Diode (SPAD); wherein the SPAD is connected to a reverse bias voltage such that the SPAD operates in geiger mode.

The pixel unit also comprises a quenching circuit unit connected with the SPAD and a pulse shaping circuit connected with the quenching circuit.

The first reading module acquires information (such as time) that the pulse signal reaches the starting point of the Hilbert linear array; and the tail reading module acquires information that the pulse signal reaches the end point of the Hilbert linear array.

The SPAD array based on the Hilbert curve further comprises a row time sequence control circuit and a column time sequence control circuit.

The working principle of the invention is as follows:

the hubert array responds to the high level of the control signal to enter a signal acquisition mode, in the acquisition mode, the SPAD senses light incident on the pixels, so that electric signals are generated, the hubert points connected with the SPAD (connected with the output of the pixels) capture the electric signals, and the electric signals are transmitted on the hubert array and transmitted to the starting point position and the end point position of the hubert array. According to the fact that the electric signals are delayed in transmission in the lead (i.e. the Hilbert linear array), the delay time of the electric signals can be obtained through testing, namely, the positions of the electric signals generated on the Hilbert linear array (i.e. the positions of the Hilbert points and the positions of the pixels which can be obtained correspondingly) can be obtained according to the time difference that the electric signals reach the first reading module and the last reading module, the positions are recorded as a number between [0:1], finally, the electric signals can be obtained in a certain row and a certain column through inverse conversion of the Hilbert curve, namely, the positions of the optical signals, the Hilbert linear array responds to the active quenching signals, and the pixels after the active quenching is completed can continuously receive photon emission pulses.

Defining a pixel corresponding to a starting point in the SPAD array based on the Hilbert curve, wherein the pixel starting point is a pixel starting point position, namely a zero point position, and marking as 0; the pixel corresponding to the endpoint is defined as the pixel endpoint position, denoted as 1. Optionally, defining the lower left corner pixel as a pixel starting point position, namely a zero point position, and recording as 0; the lower right corner pixel is defined as the pixel end position and is denoted as 1.

According to an example of the present disclosure, the optical signal is a photon, which is a fundamental particle that conveys electromagnetic interactions.

According to an example of the present disclosure, the quenching circuit unit reduces a reverse bias voltage of the SPAD such that the avalanche is quenched after the SPAD receives a single photon to generate avalanche breakdown.

According to an example of the present disclosure, the pulse shaping circuit pulse shapes the signal output by the quenching circuit and transmits the pulse signal to the head readout module and the tail readout module.

According to an example of the present patent disclosure, wherein each pixel comprises: and the time sequence control circuit generates the control signal and the active quenching clock and provides the control signal and the active quenching clock for each pixel of the SPAD array.

According to an example of the present disclosure, the first and the last readout modules are composed of three modules, namely a clock emitter, a counter and a latch, and the time when the readout pulse reaches the first and the last readout modules.

According to an example of the present disclosure, SPADs of a plurality of pixel cells are responsible for detecting and receiving photons, and hilbert linear arrays are responsible for transmitting pulse signals.

According to an example of the present patent disclosure, a pixel array composed of a plurality of pixel units and a hilbert linear array are arranged on the same layer, and an output of each pixel in the pixel array is connected to the hilbert linear array.

According to an example of the present patent disclosure, a pixel array and a hilbert array composed of a plurality of pixel units are arranged on different layers based on a 3D stack process, and outputs of individual pixels in the pixel array and corresponding hilbert lattice points on the hilbert array are electrically connected.

According to an example of the present patent disclosure, the quenching circuit and the pulse shaping circuit are integrated in the pixel cell.

The invention also provides a SPAD imaging method based on the Hilbert curve, which comprises the following steps: the pixel array (SPAD array) formed by a plurality of pixel units, the Hilbert linear array which is arranged in the form of a Hilbert curve, and Hilbert points which are connected with the output of each pixel and are positioned on the Hilbert linear array.

The Hilbert linear arrays comprise Hilbert linear arrays with different orders. Wherein the lower left corner pixel is defined as a pixel starting position, namely a zero position, which is marked as 0, and the lower right corner pixel is defined as a pixel ending position, which is marked as 1;

the hilbert linear array responds to a control signal high level to enter a signal acquisition mode, in the acquisition mode, the SPAD senses an optical signal incident on a pixel so as to generate an electric signal, a hilbert point connected with the output of the pixel captures the electric signal and transmits the electric signal on the hilbert linear array, the electric signal is transmitted to a starting point position and an end point position of the hilbert linear array, the electric signal delay time can be obtained through testing according to the transmission of the electric signal in a lead (the hilbert linear array), the electric signal can be obtained according to the time difference of the electric signal reaching a first reading module and a tail reading module so as to generate a position of the hilbert linear array (namely, the position of the hilbert point corresponds to the position of the pixel), the electric signal is recorded as a number between [0:1], finally, the electric signal can be obtained in a certain row and a certain column, namely, the position of the optical signal is converted by the hilbert linear array through the reverse conversion of the hilbert curve, and the pixel after the active quenching is finished can continuously receive photon emission pulses.

According to an example of the present disclosure, the optical signal is a photon. The imaging method further includes:

the pixel units are used for detecting photons and outputting pulse signals, the active quenching clock enables the SPAD to return to the position under the Geiger mode, the position of the photons on the Hilbert curve is obtained by remembering the time difference between the first reading module and the last reading module of the electric signal and is recorded as numbers between [0 and 1], the pulse emission conditions of different positions are reversely deduced by measuring different time differences, and the number of the pulse signals of the pixels is in direct proportion to the number of the photons, so that the number of the photons collected by each pixel can be reversely deduced, and imaging is achieved.

According to an example of the present patent disclosure, a single photon avalanche diode SPAD is employed to detect photons and output a pulse signal; wherein connecting the SPAD to a reverse bias voltage causes the SPAD to operate in geiger mode.

According to an example of the present patent disclosure, the imaging method further comprises:

the control signal is closed, the Hilbert curve SPAD array is closed, the position of photons on the Hilbert curve is obtained by remedying the time difference between the first reading module and the last reading module of the electric signal and is recorded as numbers between [0 and 1], the pulse emission conditions of different positions are reversely deduced by measuring different time differences, and the number of pulse signals of pixels is in direct proportion to the number of photons, so that the number of photons collected by each pixel can be reversely deduced, and image information can be conveniently generated based on the number of photons.

The invention also provides an image sensor comprising the SPAD array based on the hilbert curve.

The invention also provides application of the SPAD array containing the Hilbert curve in the detection field.

The invention has the beneficial effects that:

the SPAD used in the method is a single photon avalanche diode, and is a photodiode which works in a Geiger mode (reverse bias voltage is larger than avalanche breakdown voltage) and realizes single photon detection capability by utilizing avalanche breakdown. In the invention, a Hilbert linear array is adopted to fill the whole pixel array (SPAD array); the hilbert curve can linearly penetrate through each discrete unit in two dimensions or higher according to the characteristics of the self space filling curve, and only penetrates once, and each discrete unit is linearly ordered and coded, and the code is used as a unique identification of the unit; the space filling curve can map data which has no good sequence in a high-dimensional space to a one-dimensional space, and through the coding mode, objects which are adjacent in space can be stored in a block, so that IO time can be reduced, and the data processing efficiency in the memory is improved. Because the electric pulse propagates in the lead (Hilbert linear array) with time delay, the position of the photon striking the detector can be determined by only measuring the head and tail signals of the Hilbert linear array.

The invention also solves the following problems in the prior art: normally, the detector is detected by adopting a SPAD array (such as a 32×32 array), and in a general readout process, to accurately locate the position where the photon strikes the detector, readout circuits need to be designed in both the row direction and the column direction of the array, which greatly increases the overhead of the SPAD array.

Compared with the traditional SPAD array with a counting circuit, the image sensor with the single photon avalanche diode array adopts the Hilbert linear array and the corresponding imaging method, and the pixel array is greatly reduced and the pixel filling factor is higher; compared with an off-chip counting SPAD array, the ultra-high-speed reading can be realized; the invention greatly reduces the area, and is especially suitable for low-cost, low-power consumption, low-illumination and ultra-high-speed systems.

Drawings

FIG. 1 is a schematic illustration of imaging a Hilbert-curve-based SPAD array;

FIG. 2 is a schematic illustration of an array arrangement of SPAD arrays based on Hilbert curves;

fig. 3 is a circuit configuration diagram of a single pixel;

FIG. 4 is a schematic diagram of a specific configuration of a readout module;

FIG. 5 is a schematic diagram of a low order Hilbert (Hilbert) curve;

FIG. 6 is a step diagram of a Hilbert-curve-based SPAD array imaging method;

FIG. 7 is a schematic diagram of the position of a third order Hilbert in an array;

fig. 8 is an imaging method for simultaneously generating pulses for a plurality of pixel cells of a SPAD array based on a hilbert curve.

Detailed Description

The present disclosure will be described in detail below with reference to the attached drawings and the specific embodiments, which are only for explaining the principles of the present disclosure and are not intended to limit the scope of the technical solutions of the present disclosure. It will be understood that, although the terms first and second, etc. may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another element.

Fig. 1 is a schematic illustration of imaging a SPAD array based on a hilbert curve. As shown in fig. 1, the sensor is placed in front of the lens and the image of the object is mapped onto the sensor focal plane by focusing the lens when imaging. Since photons are the smallest carrier of light energy, the theoretical limit of the sensitivity of a photodetector is the ability to detect a single photon; the brighter the scene, the more photons that are passed onto the focal plane of the sensor; the darker the scene, the fewer photons will pass onto the focal plane of the sensor; the longer the imaging time, the more photons that are transmitted onto the focal plane of the sensor; the shorter the imaging time, the fewer photons that are transmitted onto the focal plane of the sensor; in practice, the exposure integration time of imaging is prolonged, which is beneficial to improving the signal-to-noise ratio.

As described above, there are various types of Image sensors such as charge coupled devices (CCDs, charge Coupled Device) and CMOS Image sensors (CIS, CMOS Image sensors), in which CCDs have advantages of high sensitivity, high-precision geometrical positions of photosensitive elements, high spatial resolution, wide dynamic response, etc., while CIS has advantages of low power consumption, high cost performance, high resolution, etc.; all that they obtain signals through sensing photons, and finally the generated Analog quantities still are voltage/current, and a complex digital-to-Analog conversion circuit (ADC) which occupies power consumption is needed, if each pixel uses one ADC module, the cost performance is too low, so that the CCD generally adopts only one ADC module, and the CIS adopts one column or one row to share one ADC module, and the cost is that the readout speed is greatly limited. The novel single photon avalanche diode (SPAD, single photon avalanchediode) of the image sensor has great advantages, the SPAD is a photodiode working under extremely high reverse bias, when single photons are incident and generate carriers, the photo-generated carriers are continuously accelerated and collide under the action of the reverse bias to generate new electron-hole pairs, so that avalanche current signals are generated in extremely short time (generally in ps magnitude), and the SPAD has single photon detection capability and can accurately measure the arrival time of the photons through the change of electric signals. The SPAD array integrated with a plurality of pixels has parallel single photon information acquisition capability, and is superior to the traditional single-point scanning architecture in detection efficiency. In addition, the photon counting circuit and the time measuring circuit can be integrated into the SPAD array, so that the functionality and the expandability of the SPAD array are obviously improved.

The invention discloses a SPAD array based on a Hilbert curve and an imaging method, and particularly relates to an image sensor with a single photon avalanche diode array by using a Hilbert linear array and a corresponding imaging method thereof.

Fig. 2 is a schematic diagram of an array arrangement of SPAD arrays based on hilbert curves according to the principles of the present patent disclosure. As shown in fig. 2, the SPAD array based on the hilbert curve includes a pixel array (SPAD array) formed by a plurality of pixel units, a hilbert linear array, a head readout module, and a tail readout module. The head reading module is connected with the starting point (lower left corner) of the Hilbert linear array, and the tail reading module is connected with the ending point (lower right corner) of the Hilbert linear array.

The Hilbert linear arrays comprise Hilbert linear arrays with different orders. Wherein the lower left corner pixel is defined as a pixel starting position, namely a zero position, which is marked as 0, and the lower right corner pixel is defined as a pixel ending position, which is marked as 1;

the hilbert linear array responds to a control signal high level to enter a signal acquisition mode, SPAD in a pixel unit senses light incident on a pixel in the acquisition mode, so that an electric signal is generated, the hilbert lattice point on the hilbert linear array connected with the output of the pixel unit captures the electric signal and transmits the electric signal to the starting point position and the end point position of the hilbert linear array through the hilbert linear array, the electric signal delay time can be obtained through testing according to the transmission of the electric signal in a lead, the electric signal can be obtained according to the time difference of the electric signal reaching a first reading module and a tail reading module, the electric signal is generated at the position of the hilbert linear array (namely the position of the hilbert lattice point) and recorded as a number between [0:1], finally, the electric signal can be obtained at a certain row of positions of the optical signal through the inverse conversion of the hilbert curve, the hilbert linear array responds to an active quenching signal, and the pixel after the active quenching is completed can continuously receive photon emission pulses.

As shown in fig. 2, the SPAD array based on the hilbert curve further includes a row timing control circuit and a column timing control circuit; wherein the row timing control circuit and the column timing control circuit are respectively connected with the rows and the columns of the pixel array. The row timing control circuit and the column timing control circuit generate an active quenching clock signal and a control signal and provide the active quenching clock signal and the control signal to the SPAD of the pixel unit. The head reading module and the tail reading module respectively receive pulse signals generated by the SPAD of the pixel unit. By the time difference of the head readout module and the tail readout module, the image information can be restored.

In accordance with an embodiment of the present patent disclosure, single photon avalanche diodes (SPADs, singlephoton avalanche diode) are employed in pixel cells to achieve hilbert curve-based imaging. Specifically, SPAD is an avalanche photodiode that operates in geiger mode. When a single photon is incident on the SPAD active region, it has a certain probability of producing a sustained avalanche breakdown. When the avalanche breakdown is quenched by the active quenching signal, the avalanche breakdown is shaped by the pulse shaping circuit and outputs a digital pulse signal to the outside. Thus, a sequence of photons incident on the SPAD in response to the output signal appears as a discrete, one-by-one pulse signal on the time axis. The number of the incident light photons, namely the light intensity, can be inverted by counting the number of the pulses, so that the gray information of the image is restored.

Specifically, fig. 3 shows a schematic circuit structure of a single pixel included in a SPAD array based on hilbert curve according to the principles of the present disclosure. As shown in fig. 3, the pixel unit includes a Single Photon Avalanche Diode (SPAD) S1, the SPAD 1 and the corresponding quenching circuit (consisting of M1, MP1, MN1 and inverter Inv 1) and the pulse shaping circuit (consisting of MP2, MN2, MP3, MN 3). Wherein the single photon avalanche diode S1 is connected to a reverse bias iHv for operation in geiger mode. The quenching circuit consists of M1, MP1 and MN1, wherein the Gate end of the M1 is connected with an input port Gate, the Gate port is connected with an active quenching clock signal, the active quenching clock signal is provided by a time sequence control circuit and a column time sequence control circuit, the drain break of the M1 is connected with the positive end of the single photon avalanche diode S1 and the Gate end of the MN1, and the source end of the M1 is connected with the substrate; the gate end of MP1 is connected with Agnd, the source end of MP1 and the substrate are connected with Avdd, the drain end of MP1 is connected with the drain end of MN1 and the input of inverter Inv 1; the gate terminal of MN1 is connected with the drain terminal of M1 and the positive terminal of single photon avalanche diode S1, the drain terminal of MN1 is connected with the drain terminal of MP1 and the input of inverter Inv1, the source terminal of MN1 and the substrate are connected with Agnd, the input of inverter Inv1 is connected with the drain terminal of MP1 and the drain terminal of MN1, and the output of inverter Inv1 is connected with output port Opulse. The pulse shaping circuit consists of MP2, MN2, MP3 and MN3, wherein the gate end of MP2 is connected with the input ports In and MN2, the source end and the substrate of MP2 are connected with the Avdd, the drain end of MP2 is connected with the drain end of MN2 and the gate end of MP3, the gate end of MN2 is connected with the input ports In and MP2, the source end and the substrate of MN2 are connected with the drain end of MP2 and the gate end of MP3 and the gate end of MN3, the gate end and the substrate of MP3 are connected with the drain end of MP2 and the drain end of MN2 and the gate end of MN3, the source end and the substrate of MP3 are connected with the drain end and the output port Out of MN3, the gate end of MN3 is connected with the drain end and the drain end of MP2 and the gate end of Out 3, and the drain end of MN3 are connected with the drain end and the output port of MP 3. The single photon avalanche diode is connected with the quenching circuit, the output Opulse of the quenching circuit is connected with In of the shaping circuit, and then Out of the shaping circuit is output.

Wherein the single photon avalanche diode S1 is connected to a reverse bias voltage iHv for operation in geiger mode. When a single photon is incident on the active region of the single photon avalanche diode S1, it has a certain probability of generating a sustained avalanche breakdown. When the avalanche breakdown is actively quenched by the quenching circuit (M1, MP1, MN1 and inverter Inv 1), the avalanche breakdown is shaped by the pulse shaping circuit (MP 2, MN2, MP3, MN 3) and outputs a digital pulse signal. Thus, when a sequence of photons is incident on the single photon avalanche diode S1, the signal output via the pulse shaping circuit appears as a discrete pulse signal one by one on the time axis.

As an example, the pixel cell includes a single photon avalanche diode SPAD, where the SPAD is connected to a reverse bias voltage such that the SPAD operates in geiger mode.

As an example, each pixel (i.e., pixel unit) further includes: and the quenching circuit is connected with the SPAD, and is used for reducing the reverse bias voltage of the SPAD to quench the avalanche after the SPAD receives a single photon to generate avalanche breakdown and outputting a pulse signal. The specific quenching process is as follows: when the SPAD receives a single photon to generate avalanche breakdown, the voltage of the positive end of the SPAD is rapidly increased, so that the potential of the Gate end of the MN1 is increased, the MN1 is conducted, the potential of the drain end of the MN1 is pulled to be equal to the potential of the Agnd, the input of the inverter is low level, the output of the inverter is high level, the Gate end of the M1 is connected with an input port Gate signal, the Gate signal receives an active quenching clock signal provided by a time sequence control circuit and a column time sequence control circuit, and when the Gate is a high level signal, the positive end of the SPAD is pulled back to be low level again, and the active quenching is completed.

As an example, each pixel further includes: and the pulse shaping circuit is connected with the quenching circuit, is used for shaping the pulse signal output by the quenching circuit and transmitting pulses to the head reading module and the tail reading module. The specific structure and operation of the head and tail read modules are shown in fig. 4.

Fig. 4 shows a specific structural schematic diagram of the readout module. As shown in fig. 4, the first readout module and the last readout module mainly comprise three modules of a clock emitter, a counter and a latch, and the main working principle is as follows: the row timing control circuit and the column timing control circuit emit control signals to the whole pixel array, the pixel units start to carry out photosensitive imaging, the single photon avalanche diode senses photons and generates avalanche effect, pulse is emitted to the Hilbert point through the quenching circuit and the shaping circuit and transmitted to the start point and the end point of the Hilbert line array through the Hilbert line array, the control signals are also applied to the counter modules of the first readout module connected with the start point of the Hilbert line array and the tail readout module connected with the end point of the Hilbert line array, the counter modules accumulate clock pulses of the clock emitter, and when the pulses on the Hilbert line array arranged according to the Hilbert curve form reach the first readout module and the tail readout module, latches in the first readout module and the tail readout module lock counter values at the moment respectively, and the difference between the counter values is the position information of the photons on the pixel array.

It should be mentioned here that the main frequency of the clock transmitters in the first readout module and the last readout module needs to be far greater than the active quenching clock signals of the line time control circuit and the column time control circuit, on one hand, the reason pulse propagates fast in the hilbert linear array, on the other hand, if the main frequency of the clock transmitters in the first readout module and the last readout module is very low, the counter still does not count the value when the pulse arrives; the counter value also needs to be fully supersampled to the pixel size of the entire array, e.g., assuming the pixel size of the array isFor a total of 1024 pixels, the counter should be at least 10 bits, otherwise the pixel cannot accurately reflect the position of the sensor photon.

Fig. 5 shows a low order Hilbert (Hilbert) curve. As shown in fig. 5, the hilbert curve is a filling curve, and similar filling curves also include Z curves, gray codes, and other methods. The hilbert curve may linearly traverse each discrete element in two or higher dimensions, and traverse each discrete element only once, depending on the characteristics of the self space-filling curve, and the linear ordering and encoding of each discrete element is performed, with the encoding being the unique identification of the element. The space filling curve can map data which has no good sequence in a high-dimensional space to a one-dimensional space, and through the coding mode, objects which are adjacent in space can be stored in a block, so that IO time can be reduced, and the data processing efficiency in the memory is improved. The Hilbert curve is oneIs passed through each point in the lattice and each point is passed only once, N satisfies the condition: />. Whereas the construction of Hilbert curve is squareThe method is a recursive process, to construct the Hilbert curves of order n, first construct 4 Hilbert curves of order n-1, the 4 Hilbert curves of order n-1 are connected by a specific order. The actual connection sequence is only 4, which is similar to the 1 st order Hilbert, but the opening direction is different. The specific operation is as follows: firstly, equally dividing a square into four small squares, starting from the square center of the southwest corner to the square center of the northwest corner, starting from the square center of the northwest corner to the square center of the northeast corner, starting from the square center of the northeast corner to the square center of the southeast corner, and starting from the square center of the southeast corner to the southeast corner, wherein the method is an iteration, if the process is continuously carried out on the four small squares, the four small squares are divided downwards, and the steps are repeated, so that a curve capable of filling the whole square is finally obtained.

In the code map of the hilbert curve, the 4 small squares divided are binary coded in clockwise order, 0.00,0.01,0.10,0.11. The latter splitting is also added with a 2-bit binary fraction based on the former code, e.g. after the second splitting of the first lattice, the resulting 4 small squares are coded as 0.0000,0.0001,0.0010,0.0011. This gives each point in the square a coding in 0,1, i.e. a one-to-one mapping from the 1 x 1 plane to the 0,1 interval is accomplished, the biggest effect of the hilbert curve is that for the N-th order hilbert curve position information between 0,1, the larger N, the more accurate the position information that is reacted.

Fig. 6 shows a step diagram of a SPAD array imaging method based on hilbert curves. As shown in fig. 6, according to the definition of the hilbert linear array, the hilbert linear array is connected to the lower left corner and the lower right corner of the pixel array (SPAD array composed of a plurality of pixel units) through the hilbert points, and is connected to each pixel in the array through the hilbert points, and each pixel passes through a curve once (i.e. is connected to only one hilbert point), so according to the expansion, 601 is the hilbert linear array, and becomes a straight linear array, the hilbert curves are expanded into a straight linear array, and each hilbert point and its corresponding SPAD (pixel unit) on the hilbert curve also sits at a position on the straight hilbert linear array along with the responsive position; if a SPAD receives a single photon to generate avalanche breakdown, reducing reverse bias voltage of the SPAD to enable the avalanche to be quenched, and outputting a pulse signal to a Hilbert linear array through a Hilbert point by a shaping circuit; because the hilbert point sits on the hilbert linear array, the pulse signal needs to propagate to two ends instead of propagating unidirectionally, and 602 shows that the pulse signal output by SPAD through avalanche quenching is output to the hilbert linear array through the shaping circuit and is propagated to the head reading module (head reading circuit) and the tail reading module (tail reading circuit) respectively. The lower left corner pixel is defined as a pixel starting point position, the leftmost position corresponding to the straight line shown by 602 is marked as 0, the lower right corner pixel is defined as a pixel end point position, and the rightmost position corresponding to the straight line shown by 602 is marked as 1; the propagation rates of the two pulses on the Hilbert linear array are the same, the pulse transmitted to the left reaches the first reading module firstly, as shown by 603, the time when the pulse reaches the first reading module is recorded, namely the time information t1 recorded by the latch in FIG. 4, and the pulse transmitted to the right still continues to be transmitted to the tail reading module; after a period of time, the pulse delivered to the right reaches the tail readout block, as shown at 604, where the latch signal in the tail readout block latches, and the photon reaches the tail readout block for a time t2.

Wherein, in addition to knowing the latch values t1 and t2 of the first and the last readout modules, the propagation speed of the pulse signal in the Hilbert linear array needs to be known, the first readout module releases a pulse slightly lower than the power supply voltage, the last readout module starts counting while releasing the pulse, and the total propagation time of the pulse in the Hilbert linear array is recordedThe position corresponding to the leftmost position of the straight line shown in fig. 6, i.e., the zero point position, is denoted as 0, and the position corresponding to the rightmost position of the straight line shown in fig. 604, i.e., the end point position, is denoted as 1; that->Representing the time from zero point to end point of the pulse signal from the Hilbert linear array and representing the whole time from 0 to 1, the speed of the pulse in the Hilbert linear array can be calculated as +.>The time of pulse reaching the first and the last readout modules is t1 and t2, t1 and t2 respectively, and the time of SPAD waiting for photons is also included, so the difference value between them is used to calculate the position, and the position at this time is assumed to be x, and x is [0:1]]Real number in between, that t1 is time information between x-0, t2 is time information between 1-x, and t2-t1 is accurate time information between 1-2x, then according to the transmission rate of pulse in Hilbert linear array->The position information at this time can be known as:

the position information is real numbers between [0,1], and according to the position information, inversion position can be carried out through the algorithm of the Hilbert curve, and MATLAB codes are as follows:

% matrix inverse transposition

function [Trans] = Transposition(Trans0)

[r,c] = size(Trans0);

for i = 1:1:r

for j = 1:1:c

Trans(i,j) = Trans0(r-j+1,c-i+1);

end

end

end

Since the core of the Hilbert curve is a connected transform, here a function defining a common inverse transpose.

clc

clear

Dim = 2;

Hilbert_old = 1;

Constant = -1;

while Constant<= Dim-2

Constant = Constant + 1 ;

a1 = Hilbert_old'+ 4^(Constant) ;

a2 = a1 + 4^(Constant) ;

a3 = Transposition(a2) +4^(Constant) ;

if mod(Constant,2) == 0

Hilbert_old = cat (1,Hilbert_old,a1);

Hilbert_h = cat(1,a3,a2);

Hilbert_old = cat (2,Hilbert_old,Hilbert_h);

else

Hilbert_old = cat (2,Hilbert_old,a1);

Hilbert_h = cat(2,a3,a2);

Hilbert_old = cat (1,Hilbert_old,Hilbert_h);

end

end

Hilbert = rot90(Hilbert_old/(2^Dim)^2);

FIG. 7 is a schematic diagram of the position of the third order Hilbert in the array, the right hand side of the array after normalization, and it can be seen that the values in the figure are all 0,1]By obtaining the time from zero to end of the pulse signal from the Hilbert linear arrayAnd the latch values t1 and t2 of the first readout module and the last readout module can obtain the position information of the hilbert point in the hilbert linear array through the calculated formula, and can know which SPAD emits pulses (namely pixel units connected with the hilbert point of the corresponding position information) through the processing algorithm and the position information inverted in fig. 7, namely the position where photons are received.

From the above, it is known that when a single SPAD (pixel cell) in a pixel array generates a pulse, the time from zero to end of the hilbert linear array is obtained by obtaining the pulse signalAnd the latch values t1 and t2 of the first readout module and the last readout module can obtain the position information of the hilbert point in the hilbert linear array according to the calculated formula, so as to obtain the position of the pixel unit, but in actual situations, in one active quenching period, a plurality of photons are emitted to the array in one active quenching period except under the condition that light rays are dark, so that the possibility that the SPADs of a plurality of pixel units are simultaneously sensitized and avalanche in a common active quenching period exists.

Fig. 8 illustrates an imaging method in which pulses are generated simultaneously by a plurality of pixel cells of a SPAD array based on a hilbert curve. As shown in fig. 8, according to the definition of the hilbert linear array, the hilbert linear array must be a curve that connects the lower left corner and the lower right corner of the pixel array and passes through each pixel in the pixel array, and each pixel passes through only once, so that according to the expansion, 801 is the hilbert linear array, and becomes a straight linear array, the hilbert curve is expanded into a straight linear array, and the SPAD array (pixel array) on the hilbert curve also sits on the straight hilbert linear array with the responding position; if a plurality of SPAD receives a single photon to generate avalanche breakdown, reducing reverse bias voltage of the SPAD to quench the avalanche, and outputting a pulse signal to a Hilbert linear array through a shaping circuit; because each Hilbert point connected with the SPAD array is located on the Hilbert linear array, pulse signals need to be transmitted to two ends instead of one-way transmission, the pulse signals which are output by a plurality of SPAD through avalanche quenching are output to the Hilbert linear array through a shaping circuit to be divided into two parts, and are respectively transmitted to a head reading circuit and a tail reading circuit, a lower left corner pixel is defined as a pixel starting point position, a leftmost position corresponding to a straight line shown by 802, namely a zero point position is marked as 0, a lower right corner pixel is defined as a pixel end point position, and a leftmost position corresponding to the straight line shown by 802, namely an end point position is marked as 1; the pulse divided into two has the same propagation speed on the Hilbert linear array, the pulse of the leftmost SPAD which senses the photon reaches the first reading module firstly, and the pulse of the rightmost SPAD which senses the photon reaches the first reading module firstlyThe first pulse reaches the tail readout module, as shown at 803, at which time the first pulse reaches the head readout module and the tail readout module, respectively, are recorded as the first readout module recorded timeAnd the time recorded by the tail reading moduleThe rest pulse signals are continuously transmitted to the head reading module and the tail reading module; after a period of time has elapsed, as shown by 804, the pulse of the SPAD sensing the photon to the next left reaches the first readout module, the pulse of the SPAD sensing the photon to the next right reaches the last readout module, and at this time, the time when the second pulse reaches the first readout module and the last readout module is recorded, which is the time +.>And the time recorded by the tail readout module +.>. Finally, the pulse of the rightmost SPAD which senses the photon finally reaches the first reading module, the pulse of the leftmost SPAD which senses the photon finally reaches the last reading module, and the time when the N-th pulse reaches the first reading module and the last reading module is recorded at the moment, wherein the time recorded by the first reading module is respectively->And the time recorded by the tail readout module +.>。

Wherein we know the time when all the pulses reach the head-to-tail pulse, and according to FIG. 8 and the above description, we know that the leftmost SPAD pulse signal of the photon sensing SPAD reaches the head-to-read module first and reaches the tail-to-read module last, thenAnd->Corresponding to t1 and t2 in fig. 6, the position information of the leftmost photon-perceived SPAD is:

the position information of the SPAD of the secondary left perceived photon is:

the position information of the rightmost photon-perceived SPAD is:

according to the position information shown in fig. 7, the position information of the pixel unit in the hilbert linear array can be obtained through the calculated formula, the position of which SPAD emits pulses, namely the position where photons are received, can be known through the processing algorithm and the position information inverted in fig. 7, the information that the SPAD receives the photons in one active quenching clock can be detected through the method, and the image information is obtained through increasing the exposure time, namely through a plurality of active quenching clocks.

Claims (6)

1. The SPAD array based on the Hilbert curve is characterized by comprising a pixel array formed by a plurality of pixel units, a Hilbert linear array, a head reading module and a tail reading module; the Hilbert linear array is a linear array which is arranged in a Hilbert curve mode, and Hilbert points are points on the Hilbert linear array; the output of each pixel is connected with a Hilbert point; the first reading module is connected with the starting point of the Hilbert linear array, and the tail reading module is connected with the end point of the Hilbert linear array; the Hilbert linear array is responsible for pulse signal transmission; the pixel unit comprises a Single Photon Avalanche Diode (SPAD); wherein the SPAD is connected to a reverse bias voltage such that the SPAD operates in geiger mode; the pixel unit further comprises a quenching circuit unit connected with the SPAD and a pulse shaping circuit connected with the quenching circuit; the first reading module acquires information that the pulse signal reaches the starting point of the Hilbert linear array; and the tail reading module acquires information that the pulse signal reaches the end point of the Hilbert linear array.

2. The SPAD array according to claim 1, wherein said SPAD array based on hilbert curve further comprises a row timing control circuit, a column timing control circuit.

3. The SPAD array according to claim 1, wherein said head and tail readout modules consist of three modules of clock transmitters, counters and latches, the time of arrival of readout pulses at the head and tail readout modules.

4. A hilbert curve-based SPAD imaging method applied to the hilbert curve-based SPAD array according to claim 1; the method comprises the steps that a Hilbert linear array responds to a control signal high level to enter a signal acquisition mode, in the acquisition mode, SPAD senses light incident on a pixel, so that an electric signal is generated, hilbert points connected with the output of the pixel capture the electric signal and transmit the electric signal to the starting point position and the end point position of the Hilbert linear array through the Hilbert linear array, the transmission of the electric signal in the Hilbert linear array has delay performance, the delay time of the electric signal can be obtained through testing, namely, the position of the Hilbert linear array, which is generated according to the time difference that the electric signal reaches a first reading module and a tail reading module, is recorded as a number between [0:1], finally, the position of the electric signal in a certain row and a certain column, namely, the position of the optical signal, can be obtained through inverse conversion of the Hilbert curve, the Hilbert linear array responds to an active quenching signal, and the pixel after the active quenching is completed continues to receive photon emission pulses.

5. The method of claim 4, wherein the imaging method further comprises: the pixel units are used for detecting photons and outputting pulse signals, the active quenching clock enables the SPAD to return to the position under the Geiger mode, the position of the photons on the Hilbert curve is obtained by remembering the time difference between the first reading module and the last reading module of the electric signal and is recorded as numbers between [0 and 1], the pulse emission conditions of different positions are reversely deduced by measuring different time differences, and the number of the pulse signals of the pixels is in direct proportion to the number of the photons, so that the number of the photons collected by each pixel can be reversely deduced, and imaging is achieved.

6. An image sensor comprising a SPAD array according to any of claims 1-3 based on a hilbert curve.