Event-driven wide dynamic range image sensor and process

The pixel circuit with multiple reading phases using voltage ramps addresses the challenge of achieving a wide dynamic range in CMOS image sensors, enhancing image quality by preventing saturation and charge migration.

FR3131494B1Active Publication Date: 2024-11-08COMMISSARIAT A LENERGIE ATOMIQUE ET AUX ENERGIES ALTERNATIVES
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
FR2021014600
Authority / Receiving Office
FR · FR
Patent Type
Patents
Current Assignee / Owner
Filing Date
2021-12-29
Publication Date
2024-11-08
Estimated Expiration
2041-12-29

AI Technical Summary

Technical Problem

CMOS image sensors with smaller pixels face challenges in achieving a wide dynamic range due to lower charge storage capacity, leading to pixel saturation and undesirable image degradation, particularly in high-illuminance regions, and existing solutions either increase complexity or suffer from alignment issues and power consumption.

Method used

A pixel circuit with a pinched photodiode and a comparator that performs multiple reading phases using voltage ramps to store pixel data in a memory, allowing for a wide dynamic range without increasing pixel circuit area or power consumption.

Benefits of technology

The solution enables a wide dynamic range image sensor with improved image quality by effectively handling varying illumination levels without increasing complexity or power consumption, preventing pixel saturation and charge migration.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 00000039_0000
    Figure 00000039_0000
  • Figure 00000039_0001
    Figure 00000039_0001
  • Figure 00000040_0000
    Figure 00000040_0000
Patent Text Reader

Abstract

Event-driven method and wide dynamic range image sensor. This description relates to a pixel circuit of an image sensor, the pixel circuit comprising: a pinch-off photodiode (PD) connected to a readout node (SN); a comparator (104) configured: to compare, during a first readout phase, a first readout node voltage (Vpix) at the readout node (SN) with a first voltage ramp (V_RAMP) and to generate an event signal (EVENT) if the voltage of the first voltage ramp (V_RAMP) crosses the first readout node voltage (Vpix); and, otherwise, to compare, during a second readout phase, a second readout node voltage (Vpix) at the readout node (SN) with a second voltage ramp (V_RAMP) and to generate an event signal (EVENT) when the voltage of the second voltage ramp (V_RAMP) crosses the second readout node voltage (Vpix);and an event signaling circuit (1102) configured to signal the generation of the event signal during the first or second read phase to an event recording circuit. Figure for the abstract: Fig. 11;
Need to check novelty before this filing date? Find Prior Art

Description

Title of the invention: Event-driven wide dynamic range image sensor and method Technical field

[0001] The present description relates generally to the field of image sensors and, in particular, to CMOS image sensors having pinched photodiodes. Prior art

[0002] There is a continuous progress in the field of CMOS image sensors towards smaller and smaller pixels, particularly with the widely adopted 3D stacking integration techniques. Smaller pixels mean smaller photodiodes, which in turn lead to a lower capacity to store photo-generated charges. The result is that sensor saturation is reached at lower illumination levels, and the dynamic range is thus restricted.

[0003] Providing image sensors with a wide dynamic range is particularly important for achieving satisfactory image quality. This is because in a typical scene, there is often considerable variation in illumination levels from the brightest to the darkest regions. A low dynamic range can cause saturation of pixels in high-illumination image regions, which is not only disadvantageous for the affected pixels, but can also lead to the migration of photo-generated charges from the saturated pixels to the surrounding pixels, leading to undesirable degradation of image quality.

[0004] Although solutions have been proposed in the past to improve the dynamic range of CMOS image sensors, these solutions tend to fall into two categories: 1) solutions that modify the pixel structure, generally leading to greater complexity and / or increased pixel circuitry area; or 2) solutions based on multiple exposures that may suffer from drawbacks such as residual image effects and misalignment problems in the case of acquisitions that are not correlated in time, as well as the need for readout with a post-production processing operation that costs chip area and leads to increased power consumption. Summary of the Invention

[0005] An object of the embodiments of the present description is to overcome all or part of the drawbacks of the prior art.

[0006] According to one aspect, a pixel circuit of an image sensor is provided, the pixel circuit comprising: a pinched photodiode connected to a readout node; a comparator configured: to compare, during a first read phase, a first read node voltage at the read node, resulting from a first integration period of the pinched photodiode, with a first voltage ramp and to generate an output signal if the voltage of the first voltage ramp crosses the first read node voltage; and, if the voltage of the first voltage ramp does not cross the first read node voltage during the first read phase, to compare, during a second read phase, a second read node voltage at the read node, resulting from the first integration period and a second integration period of the pinched photodiode, with a second voltage ramp and to generate an output signal when the voltage of the second voltage ramp crosses the second read node voltage;and a memory configured to receive a digital ramp and to store, in response to the output signal of the comparator, during the first or second reading phase, a value of the digital ramp to constitute pixel data of the pixel.;

[0007] According to one embodiment, the pixel circuit further comprises a logic circuit configured to deactivate the comparator during the second reading phase if the output signal is generated during the first reading phase.

[0008] According to one embodiment, the second ramp signal has an amplitude greater than the amplitude of the first ramp signal.

[0009] According to one embodiment, the memory is further configured to receive a digital signal having a first value during the first reading phase and a second value during the second reading phase, the memory being configured to store, in response to the output signal of the comparator during the first or second reading phase, the value of the digital signal.

[0010] According to one embodiment, a cathode of the photodiode is connected to the reading node via a transfer gate, the transfer gate being activated by a first activation pulse at the start of the first reading phase and by a second activation pulse at the start of the second reading phase, the voltage at the reading node not being reset between the first and second reading phases.

[0011] According to one embodiment, the pixel circuit further comprises a reset transistor connecting the cathode of the photodiode to a reset voltage rail and the logic circuit is further configured to activate the reset transistor in response to the output signal of the comparator during the first or second reading phase.

[0012] According to one embodiment: if the voltage of the second voltage ramp does not cross the second reading node voltage during the second reading phase, the comparator is further configured to compare, during a third read phase, a third read node voltage at the read node with a third voltage ramp and to generate an output signal when the voltage of the third voltage ramp crosses the third read node voltage; and the memory is further configured to store, in response to the output signal of the comparator during the third read phase, a value of the digital ramp to constitute pixel data of the pixel.

[0013] According to another aspect, an image sensor is provided comprising an array consisting of a plurality of the foregoing pixel circuit.

[0014] According to one embodiment, the first voltage ramp crosses the first read node voltage of a first of the pixel circuits during the first read phase and the memory of the first pixel circuit is configured to store: a first value of the digital ramp corresponding to the first read node voltage, and the first value of the digital signal; and the second voltage ramp crosses the second read node voltage of a second of the pixel circuits during the second read phase and the memory of the second pixel circuit is configured to store: a second value of the digital ramp corresponding to the second read node voltage, and the second value of the digital signal.

[0015] According to one embodiment, the image sensor further comprises an image reconstruction device configured to reconstruct an image by generating: a first pixel value of the image based on the first values ​​stored in the memory of the first pixel circuit; and a second pixel value of the image based on the second values ​​stored in the memory of the second pixel circuit and based on the relative durations of the first and second integration periods.

[0016] According to one embodiment, the image sensor further comprises: a counter configured to generate the digital ramp and to provide the digital ramp to each of the plurality of pixel circuits; and a digital to analog converter configured to convert the digital ramp to the first voltage ramp during the first read phase and to the second voltage ramp during the second read phase.

[0017] According to one embodiment, the image sensor further comprises a reading interface configured to read, after the first and second reading phases, an image from the image sensor by reading the pixel data from the memory of each pixel circuit.

[0018] According to one embodiment, the image sensor comprises stacked first and second stages, the pixel circuit array being formed in the first stage and the comparator being formed in the second stage.

[0019] According to one embodiment, the memory is formed in the second stage, or in a third stage of the image sensor stacked with the first and second stages.

[0020] According to another aspect, a method of reading a pixel circuit of an image sensor is provided, the pixel circuit comprising a pinched photodiode connected to a reading node, the method comprising: comparing, by a comparator during a first reading phase, a first reading node voltage at the reading node, resulting from a first integration period of the pinched photodiode, with a first voltage ramp and generating, by the comparator, an output signal if the voltage of the first voltage ramp crosses the first reading node voltage;if the voltage of the first voltage ramp does not cross the first read node voltage during the first read phase, comparing, by the comparator, during a second read phase, a second read node voltage at the read node, resulting from the first integration period and a second integration period of the pinched photodiode, with a second voltage ramp and generating an output signal when the voltage of the second voltage ramp crosses the second read node voltage; receiving, by a memory of the pixel circuit, a digital ramp; and storing, in response to the output signal of the comparator, during the first or second read phase, a value of the digital ramp to constitute pixel data of the pixel. ;

[0021] According to another aspect, a pixel circuit of an image sensor is provided, the pixel circuit comprising: a pinched photodiode connected to a read node; a comparator configured: to compare, during a first read phase, a first read node voltage at the read node, resulting from a first integration period of the pinched photodiode, with a first voltage ramp and to generate an event signal if the voltage of the first voltage ramp crosses the first read node voltage;and, if the voltage of the first voltage ramp does not cross the first read node voltage during the first read phase, to compare, during a second read phase, a second read node voltage at the read node, resulting from the first integration period and a second integration period of the pinched photodiode, with a second voltage ramp and to generate an event signal when the voltage of the second voltage ramp crosses the second read node voltage; and an event signaling circuit configured to signal the generation of the event signal during the first or second read phase to an event recording circuit. ;

[0022] According to one embodiment, the event signaling circuit is further configured to deactivate the comparator during the second reading phase if the event signal is generated during the first reading phase.

[0023] According to one embodiment, the second signal ramp has a greater amplitude larger than the amplitude of the first signal ramp.

[0024] According to one embodiment, a cathode of the photodiode is connected to the reading node via a transfer gate, the transfer gate being activated by a first activation pulse at the start of the first reading phase and by a second activation pulse at the start of the second reading phase, the voltage at the reading node not being reset between the first and second reading phases.

[0025] According to one embodiment, the pixel circuit further comprises a reset transistor connecting the cathode of the photodiode to a reset voltage rail and the event signaling circuit is further configured to activate the reset transistor in response to the generation of the event signal by the comparator during the first or second read phase.

[0026] According to one embodiment, the pixel circuit further comprises a logic device connected to the output of the comparator and configured to deactivate the transfer gate in response to the generation of the event signal during the first or second reading phase.

[0027] According to another aspect, there is provided an image sensor comprising: an array consisting of a plurality of the preceding pixel circuit; a counter configured to generate the digital ramp and to provide the digital ramp to an image memory; and a digital to analog converter configured to convert the digital ramp to the first voltage ramp during the first read phase and to the second voltage ramp during the second read phase.

[0028] According to one embodiment, the image sensor further comprises an image reconstruction device configured to reconstruct an image by generating: a first pixel value of the image based on a first value of the digital ramp associated with a first circuit of the plurality of pixel circuits; and a second pixel value of the image based on a second value of the digital ramp associated with a second circuit of the plurality of pixel circuits and based on the relative durations of the first and second integration periods.

[0029] According to one embodiment, the image sensor further comprises a control circuit configured, during the first and second reading phases, to interrupt the incrementation of the counter at each count value of the digital ramp until a corresponding end-of-scan signal is activated by the event recording circuit indicating that all the pixel circuits generating the event signal corresponding to the count value have been read.

[0030] According to one embodiment, the control circuit is configured to delay a start of the second read phase until all pixel circuits having events have been read during the first read phase.

[0031] According to one embodiment, the control circuit is configured to perform the first reading phase in two parts, a first part after a first integration time and a second part after a second integration time.

[0032] According to one embodiment, the control circuit is configured to generate an image histogram by counting the number of pixel circuits signaling events during each count value of the counter.

[0033] According to one embodiment, the image memory is further configured to receive a digital signal having a first value during the first reading phase and a second value during the second reading phase, the image memory being configured to store, in response to the event signal from one of the pixel circuits during the first or second reading phase, the value of the digital signal.

[0034] According to one embodiment, each pixel circuit further comprises a register connected to the output of the comparator and configured to store, in response to the generation of the event signal, a digital signal having a first value during the first reading phase and a second value during the second reading phase.

[0035] According to one embodiment, the image sensor comprises stacked first and second stages, the pixel circuit array being formed in the first stage and the comparator being formed in the second stage.

[0036] According to one embodiment, the event signaling circuit is formed in the second stage, or in a third stage of the image sensor stacked with the first and second stages.

[0037] According to another aspect, a method of reading a pixel circuit of an image sensor is provided, the pixel circuit comprising a pinched photodiode connected to a read node, the method comprising: comparing, by a comparator during a first read phase, a first read node voltage at the read node, resulting from a first integration period of the pinched photodiode, with a first voltage ramp and generating, by the comparator, an event signal if the voltage of the first voltage ramp crosses the first read node voltage;if the voltage of the first voltage ramp does not cross the first read node voltage during the first read phase, comparing, by the comparator during a second read phase, a second read node voltage at the read node, resulting from the first integration period and a second integration period of the pinched photodiode, with a second voltage ramp and generating an event signal when the voltage of the second voltage ramp crosses the second read node voltage; and signaling, by an event signaling circuit, the generation; of the event signal during the first or second reading phase to an event recording circuit.

[0038] According to one embodiment, the method further comprises: receiving, by an image memory of the image sensor, a digital ramp; and storing, at an address of the image memory associated with the pixel circuit and in response to the event signal of the comparator during the first or second read phase, a value of the digital ramp to constitute pixel data of the pixel. Brief description of the drawings

[0039] These characteristics and advantages, as well as others, will be explained in detail in the following description of particular embodiments given without limitation in relation to the attached figures among which:

[0040] [Fig.l] schematically represents a pixel circuit according to an exemplary embodiment of the present description;

[0041] [Fig.2] schematically represents the pixel circuit of [Fig.l] in more detail according to an exemplary embodiment of the present description;

[0042] [Fig.3] schematically represents the pixel circuit of [Fig.1] in more detail according to another exemplary embodiment of the present description;

[0043] [Fig.4] schematically represents a pixel circuit according to another exemplary embodiment of the present description;

[0044] [Fig.5] schematically represents an image sensor comprising an array of pixel circuits according to an exemplary embodiment of the present description;

[0045] [Fig.6] is a timing diagram representing reading phases of the image sensor of [Fig.5] according to an exemplary embodiment of the present description;

[0046] [Fig.7] is another timing diagram representing reading phases of the image sensor of [Fig.5] according to an exemplary embodiment of the present description;

[0047] [Fig.8] is a graph representing digital pixel values ​​captured by pixels of the image sensor of [Fig.5] according to an exemplary embodiment of the present disclosure;

[0048] [Fig.9] is a graph representing digital pixel values ​​captured by pixels of the image sensor of [Fig.5] as a function of brightness according to another exemplary embodiment of the present disclosure;

[0049] [Fig. 10] is a graph representing a variable step height of a ramp signal according to an exemplary embodiment of the present disclosure;

[0050] [Fig.l 1] schematically represents a pixel circuit according to another exemplary embodiment of the present description;

[0051] [Fig. 12] schematically represents the pixel circuit of [Fig. 11] in more detail according to another exemplary embodiment of the present description;

[0052] [Fig. 13] schematically represents a pixel circuit according to yet another exemplary embodiment of the present description;

[0053] [Fig. 14] schematically represents an image sensor comprising an array of pixel circuits according to another exemplary embodiment of the present description;

[0054] [Fig. 15] is a timing diagram representing reading phases of the image sensor of [Fig. 14] according to an exemplary embodiment of the present description;

[0055] [Fig. 16] is another timing diagram representing reading phases of the image sensor of [Fig. 14] according to an exemplary embodiment of the present description;

[0056] [Fig. 17] is another timing diagram representing reading phases of the image sensor of [Fig. 14] according to another exemplary embodiment of the present description;

[0057] [Fig. 18] is another timing diagram representing reading phases of the image sensor of [Fig. 14] according to yet another exemplary embodiment of the present description;

[0058] [Fig. 19] schematically represents a pixel circuit according to yet another exemplary embodiment of the present disclosure; and

[0059] [Fig.20] is a timing diagram representing reading phases of the image sensor of [Fig. 14] according to yet another exemplary embodiment of the present description. Description of the embodiments

[0060] The same elements have been designated by the same references in the different figures. In particular, the structural and / or functional elements common to the different embodiments may have the same references and may have identical structural, dimensional and material properties.

[0061] Unless otherwise specified, when referring to two elements connected to each other, this means directly connected without intermediate elements other than conductors, and when referring to two elements connected (in English "coupled") to each other, this means that these two elements can be connected or be connected by means of one or more other elements.

[0062] In the following description, when referring to absolute position qualifiers, such as the terms "front", "back", "top", "bottom", "left", "right", etc., or relative position qualifiers, such as the terms "above", "below", "upper", "lower", etc., or to orientation qualifiers, such as the terms "horizontal", "vertical", etc., it is made reference unless otherwise specified to the orientation of the figures.

[0063] Unless otherwise specified, the expressions "about", "approximately", "substantially", and "of the order of" mean to within 10%, preferably to within 5%.

[0064] First aspect - Pixel circuit with local memory

[0065] [Fig.l] schematically represents a pixel circuit 100 according to an exemplary embodiment of the present description. The pixel circuit 100 comprises a sensor 102 and a reading circuit 103 comprising a comparator (COMP) 104.

[0066] The sensor 102 comprises a photodiode PD, which is for example a pinched photodiode, having its anode connected to a ground rail of the sensor 102 and its cathode connected to a reading node SN of the sensor 102, for example via a transfer gate 106. The transfer gate 106 is, for example, controlled by a control signal TC.

[0067] In some embodiments, the sensor 102 further comprises a reset transistor MRST connecting the cathode of the photodiode PD to a reset voltage rail Vrst. For example, the reset transistor is a MOS transistor having its source connected to the reset voltage rail Vrst. The photodiode PD has an associated capacitance CPD between its cathode and its anode. The reading node SN also has an associated capacitance CSN with respect to the ground rail of the sensor 102, this capacitance CSN resulting for example from parasitic capacitances and / or a capacitor (not shown) connected between the reading node SN and the ground rail.

[0068] The reading circuit 103 of the pixel circuit 100 comprises, for example, in addition to the comparator 104, a memory (MEM) 108 and a logic circuit (LOGIC) 110.

[0069] The comparator 104 has, for example, a negative input connected to the read node SN and a positive input connected to an input line of the pixel circuit receiving a voltage ramp V_RAMP. An output of the comparator 104 provides a signal WRITE when the voltage ramp V_RAMP crosses a voltage Vpix present at the read node SN. The output of the comparator is, for example, connected to a write activation input of the memory 108, to an input of the logic circuit 110 and, via an automatic zeroing switch 112, to the negative input of the comparator 104. The automatic zeroing switch 112 is controlled by an automatic zeroing control signal AZ.

[0070] The logic circuit 110 receives, for example, an initialization signal INIT and is configured, in response to this command, to trigger the logic circuit 110 in order to, for example, restore the functionalities of the comparator 104 and / or the photodiode PD by bringing a deactivation signal DIS AB LE to a non-activated state. The deactivation signal DISABLE is, for example, supplied to an enable input of the comparator 104 and / or is used to activate the reset transistor MRST, if present.

[0071] The memory 108 comprises for example a first part 114 configured to receive a digital ramp signal DATA_RAMP, which is an 8-bit signal in the example of [Fig.l], but could have a different number of bits. The memory 108 also comprises for example a second part 116 configured to receive a digital stamp value STAMP_INT, which is a 2-bit signal in the example of [Fig.l], but could have a different number of bits. The memory 108 also receives a read command signal READ on an input line and provides on an output line, in response to the read command READ, a pixel data value DATA_PIX comprising data values ​​held in the first and second memory portions 114, 116. In the example of [Fig.l], the pixel data value DATA_PIX is a 10-bit value, although in alternative embodiments it could have a different number of bits.

[0072] In operation, the pixel circuit 100 is capable of performing multiple readout phases in order to obtain a relatively wide dynamic range. The deactivation signal DISABLE is for example initially in an activated state triggered by the activation of the WRITE signal at the output of the comparator during a previous cycle. This implies that the reset transistor MRST is activated, and the voltage at the cathode of the photodiode PD is at the reset voltage Vrst. The voltage of the readout node Vpix is ​​initialized to an initial value of the ramp voltage V_RAMP, equal for example to the reset voltage Vrst, or to a different voltage, by applying the signal AZ. The transfer gate control signal TG is also activated for example in order to initialize the photodiode PD.The INIT signal is then, for example, activated so that the logic circuit 110 deactivates the DISABLE deactivation signal and the TG control is brought to the low state, causing the start of an integration period.

[0073] After a first integration period, a first reading phase is triggered by applying a pulse to the transfer gate control signal TG. This causes charges accumulated by the photodiode PD during the first integration period to be transferred to the reading node SN, thereby reducing the reading node voltage as a function of the illumination of the photodiode during the first integration period. Indeed, it is assumed in the present description that the photodiode is of a type whose transfer of charges to the reading node is subtractive, i.e. due to a transfer of electrons and, consequently, the reset voltage Vrst applied to the reading node SN is in the high state and is reduced by an amount which is proportional to the amount of accumulated charges.In alternative embodiments, it would also be possible for the photodiode to be of a type such that the charge transfer to the read node SN is additive, i.e. due to hole transfer, in which case the reset voltage Vrst is . a low-state voltage that is increased by an amount that is proportional to the amount of accumulated charge.

[0074] The voltage Vpix at the read node SN is then compared, by the comparator 104, to a first voltage ramp of the signal V_RAMP. The first voltage ramp has for example decreased in amplitude compared to the full range between the lowest saturation voltage Vsat of the pixel circuit and the reset voltage level Vrst. For example, the first voltage ramp extends from Vsat to Vsat*M, where M is for example equal to between 0.2 and 0.8, and for example to about 0.5. Thus, the conversion time, corresponding to the rise time of the ramp, is reduced compared to a full voltage ramp.

[0075] If the voltage ramp V_RAMP exceeds the value of the read node voltage Vpix, the WRITE signal at the output of the comparator 104 is activated. This causes, for example, the memory 108 to store current values ​​of the digital signals DATA_RAMP and STAMP_INT in the first and second parts 114, 116, respectively. Furthermore, the WRITE signal causes, for example, the logic circuit 110 to activate the disable signal, thereby disabling the comparator 104 and / or activating the reset transistor MRST to disable the photodiode PD until a subsequent image capture operation, when the INIT command is reactivated.

[0076] According to a variant, if the voltage ramp V_RAMP does not exceed the value of the voltage Vpix, the WRITE signal at the output of the comparator 104 is not activated and the integration by the photodiode PD continues. The charges previously transferred are thus maintained at the node SN at Vpix

[0077] One or more additional read phases can then be performed in a similar way to the first read phase, also with voltage ramps of reduced amplitudes compared to the full range. However, a final read phase is for example performed with a full amplitude ramp, in other words, a ramp rising from Vsat to Vrt, which ensures that all pixel circuits of the array are read during at least one of the read phases. No reset is performed between one read phase and the next of the same read cycle, and it will be noted that, in the case of a pinched photodiode, the application of the TG pulse is cumulative, and not destructive. Each phase is associated with a different integration time and it is thus possible to obtain a wide dynamic range read.Indeed, for pixels with high illumination, the voltage of the read node Vpix will be relatively low and can be read during the first reading phase. Pixels receiving lower illuminations will lead to smaller drops in the voltage of the read node Vpix, and are read during a later reading phase, associated with a longer integration time.

[0078] [Fig.2] schematically represents the pixel circuit 100 of [Fig.l] more in detail according to an exemplary embodiment of the present description. The sensor 102 is the same as in the example of [Fig.l], except that the reset transistor MRST is implemented as a P-channel MOS transistor (PMOS).

[0079] In the example of Figure 2, the comparator 104 comprises a first stage implemented by N-channel MOS (NMOS) transistors M1, M2, MSTBY and MBIAS, and by PMOS transistors M3 and M4. The transistors M1 and M2 constitute a differential pair, the gate of the transistor M1 being connected to the negative input of the comparator 104 and the gate of the transistor M2 being connected to the positive input of the comparator 104. One of the main current-conducting nodes, for example the sources, of each of the transistors M1 and M2 are connected, or linked, to a common node 202 which is in turn connected via the bias transistor MBIAS and the bias transistor MSTBY to the ground rail.The gate of the bias transistor MBIAS receives for example a bias voltage V_BIAS and the gate of the standby transistor MSTBY is for example connected to the deactivation line which provides the deactivation signal which, in figure 2, is an inverted deactivation signal D1SÀBLE-The other main current conduction node, for example the drain, of the transistor M1 provides for example the output voltage of the first stage of the comparator 104 and is for example connected via the transistor M3 to the supply rail Vdd. The gates of the transistors M3 and M4 are for example connected to each other and to the drain of the transistor M4, which is for example a common drain node of the transistors M4 and M2. The automatic zeroing switch 112 is for example implemented by an NMOS transistor MAZ which connects, via its main current conduction nodes, the gate and the drain of the transistor ML.

[0080] The comparator 104 further comprises a second stage 204 comprising a Schmitt trigger 206 which generates for example the output signal WRITE of the comparator 104. The Schmitt trigger 206 introduces a hysteresis in order to compensate for small transitions by the first stage of the comparator and also ensures immunity against noise.

[0081] The memory 108 is for example implemented by an array, possibly a linear array, of SRAM cells (from the English "static random-access memory"). In certain embodiments, the memory 108 is a two-port memory, so that simultaneous write and read operations are possible, making it possible for example to read the memories 108 of the pixels of the array at the same time as the integration and the ADC ramp conversion are carried out.

[0082] In the embodiment of [Fig.2], the pixel circuit 100 is implemented in three stages TIER 1, TIER 2 and TIER 3 of the three-dimensional stack. The stages TIER 1, TIER 2 and TIER 3 constitute for example a sensor level (SENSOR LEVEE) and are for example constituted by a sequential stack. The TIER 3 stage constitutes, for example, a logic level (LOGIC LEVEL) such as an advanced logic level. The TIER 1 stage comprises, for example, the sensor 102, and the TIER 2 stage comprises, for example, the first stage of the comparator and the automatic zeroing transistor MAZ.

[0083] An interconnection 208 between the reading node SN of the sensor 102 and the negative input of the comparator 104 intended to transmit the reading node voltage Vpix is ​​for example implemented by a via between the stages TIER 1 and TIER 2. Similarly, an interconnection 210 between the deactivation line provided by the logic circuit 110 and the gate of the reset transistor MRST of the sensor 102 is for example implemented by an additional via between the stages TIER 1 and TIER 2.

[0084] An interconnection 211 between the output of the first stage of the comparator and the input of the Schmitt trigger 206 is for example implemented using hybrid connection contacts between the stages TIER 2 and TIER 3. Similarly, an interconnection 212 between the deactivation output of the logic circuit 110 and the gate of the transistor MSTBY and the via 210 is for example also implemented using hybrid connection contacts between the stages TIER 2 and TIER 3.

[0085] [Fig. 3] schematically represents the pixel circuit 100 of [Fig. 1] in more detail according to another exemplary embodiment of the present description. The example of [Fig. 3] is similar to that of [Fig. 2], except that the transistors M3 and M4 of the first stage of the comparator are moved into the TIER 3 stage. Such a solution can be advantageous in the case where space is limited in the TIER 2 stage. There is therefore an additional interconnection 302 between the TIER 2 and TIER 3 stages in order to connect the transistors M2 and M4, this interconnection being for example implemented by a hybrid connection contact.

[0086] [Fig. 4] schematically represents a pixel circuit 400 according to another exemplary embodiment of the present disclosure. The pixel circuit 400 is similar to the pixel circuit 100 of [Fig. 3], except that modifications have been made to reduce the interconnections between the stages. Indeed, it may be desirable to limit, in particular, the number of hybrid connection contacts between the TIER 2 and TIER 3 stages. Consequently, in the embodiment of [Fig.4], the reset transistor MRST of the sensor 102 is no longer present, thus making it possible to remove the interconnection 210 between the TIER 1 and TIER 2 stages. Furthermore, the standby transistor MSTBY is removed from the TIER 2 stage and instead placed in the TIER 3 stage between a common node 402 connected, or connected, to the sources of the transistors M3 and M4, and the supply rail Vdd.The standby transistor MSTBY is also implemented for example in the form of a PMOS transistor rather than an NMOS transistor and the deactivation signal DIS AB LE generated by the logic circuit 110 and applied to the gate of the standby transistor MSTBY is therefore, in this embodiment, no longer a . inverted signal. So there are only two interconnections 302 and 211 between the TIER 2 and TIER 3 stages.

[0087] [Fig.5] schematically represents an image sensor 500 comprising a network of pixel circuits according to an exemplary embodiment of the present disclosure. The pixel circuits are for example implemented by the pixel circuit 100 of Figures 1, 2 or 3, or by the pixel circuit 400 of [Fig.4]. In particular, the image sensor 500 comprises for example a sensor array (SENSOR ARRAY) 502 comprising, for example, one or more stages which comprise the sensors 102 and at least a portion of the comparators of the pixel circuits, and a logic array (LOGIC ARRAY) 504 stacked on the sensor array 502 and comprising, for example, the memory 108 and a logic circuit 110 of the pixel circuits and, in some cases, a portion of the comparators 104 of the pixel circuits.

[0088] The image sensor 500 also comprises, for example, a counter (COUNTER) 506, which is, for example, an 8-bit, 10-bit or 12-bit counter, configured to generate the digital ramp signal DATA_RAMP which is supplied to the logic network 504 and, in particular, to the memory 108 of each of the pixel circuits. Furthermore, the image sensor 500 comprises, for example, a digital-to-analog converter (DAC) 508 configured to convert the digital ramp signal DATA_RAMP into a voltage signal in order to generate the voltage ramp V_RAMP supplied to the sensor network 502 and, in particular, to the comparator 104 of each of the pixel circuits.

[0089] The image sensor 500 also comprises, for example, a control circuit (CONTROL) 510, implemented, for example, by a finite state machine and configured to generate the signals STAMP_INT and INIT, which are supplied to the logic network 504 and, in particular, to the memory 108 and to the logic circuit 110, respectively, of each pixel circuit and the signals TG, AZ and V_BIAS, which are supplied to the sensor network 502 and, in particular, to the sensor 102, to the automatic zeroing switch 112 and to the comparator 104 of each pixel circuit.

[0090] The image sensor 500 also comprises, for example, a memory address decoder (MEMORY ADDRESS DECODER) 512 connected to the logic network 504 and an output interface (O / P INTERFACE) 514 connected to the memory address decoder 512. The output interface 514 is, for example, configured to read images captured by the sensor network by providing the read signal READ and an address signal ADDRESS to the memory address decoder 512. There is, for example, a pixel data bus going from the memory 108 of each pixel circuit to the memory address decoder 512 and the address signal ADDRESS designates, for example, a single pixel circuit which must be read. The memory address decoder 512 is, for example, configured to provide the signal READ to the memory 108 of the pixel circuit designated by the address and the memory 108 of the pixel circuit responds by providing the corresponding pixel data signal DATA_PIX from its memory.

[0091] The image sensor 500, or an external circuit (not shown), further comprises, for example, an image reconstruction device (HDR IMAGE RECONST.) 516 which is, for example, configured to reconstruct a wide dynamic range image based on pixel measurements from the array, as will be described in more detail later. For example, the device 516 is implemented by a processing device, such as a CPU (central processing unit) or an ISP (image signal processor).

[0092] [Fig. 6] is a timing diagram showing readout phases of the image sensor 500 of [Fig. 5] according to an exemplary embodiment of the present disclosure. The example of [Fig. 6] is based on four readout phases, but in alternative embodiments there could be at least two readout phases and, for example, up to 6, 8, 12, 16 or 32 readout phases. [Fig. 6] shows the time (TIME) along the x-axis and the pixel value (PIXEL VALUE) along the y-axis, where the pixel value corresponds, for example, to the pixel voltage Vpix. This pixel voltage is for example in a range from a low voltage level Vsat, for example at or close to the ground voltage (0 V) corresponding to the lowest voltage to which the read node SN can drop when the illumination is at a maximum and a high voltage level corresponding to the reset voltage Vrst.

[0093] An integration period for an n-th frame (FRAME_n) begins at a time To.

[0094] A first reading phase begins at a time T! after the TG command ai been pulsed, this TG pulse being for example a global command for all pixels in the array. All photo-generated carriers during the period from To to 7\ are simultaneously transferred and stored at the corresponding Vpix nodes. The first readout phase involves a voltage ramp V_RAMP having a range from Vsat to Vm, where Vm is for example the midpoint voltage between the Vsat and Vrst voltages. In the case where the charge transfer is subtractive as in the example of [Fig.6] and assuming that Vsat is the ground voltage, the voltage Vm is chosen for example as Vrst / 2. Conversely, in the case (not shown) where the charge transfer is additive, the Vrst and Vsat voltages are inverted and if the Vrst voltage is at the ground voltage, the Vm voltage is chosen for example as Vsat / 2.

[0095] A line L0 in Figure 6 represents a pixel having a read node voltage Vpix falling from Vrst at time 70 to Vsat at time T ï which corresponds to a maximum amount of illumination without exceeding the saturation limit. The integration time T\ is for example chosen so that no pixel saturates before the time Tj is reached. During the first reading phase, all pixels having read node voltages between the voltages Vsat and Vm will trigger the corresponding comparator and thus the value will be converted and stored in the memory 108. These pixels correspond to those having read node voltages Vpix between the lines LO and L1 in Figure 6, where L1 represents a pixel having a read node voltage Vpix falling from Vrst at the time Tq to Vm at the time Tb. These pixels have been read and, therefore, they are for example deactivated for the remainder of the network reading cycle, in other words until the start of the next frame capture. All other pixels retain for example the stored value Vpix, which for these pixels is between the voltages Vm and Vrst.During the first reading phase, the stamp signal STAMP_INT has for example a first binary value for example equal to "11" binary.

[0096] A second reading phase begins at a time T2 after the TG command has been pulsed again, this TG pulse being for example a global command for all the pixels of the array. In the unread pixels, in other words the pixels which are not read during the first reading phase, the voltage Vpix stored in the first reading phase is accumulated with the new photo-generated carriers between the period from T s to C2. The second reading phase involves for example the same voltage ramp from Vsat to Vm as that applied during the first reading phase.The integration time T 2 is for example chosen to correspond to twice T ( which is a time at which pixels having a read node voltage just above Vm at time T ! have a read node voltage Vpix dropping to Vsat at time T2, as represented by line LL During the second read phase, all pixels having read node voltages Vpix between voltages Vsat and Vm will trigger their corresponding comparator and thus be read and deactivated, as for the pixels read during the first read phase. These pixels correspond to those having read node voltages Vpix between lines L1 and L2 in Figure 6, where L2 represents a pixel having a read node voltage Vpix dropping to Vm at time T2, All other pixels retain for example the stored value Vpix, which for these pixels is between voltages Vm and Vrst.During the second reading phase, the stamp signal STAMP_INT has for example a second binary value different from the first and equal for example to binary "10".

[0097] A third reading phase begins at a time T3 after the TG command has been pulsed again, this TG pulse being for example a global command for all the pixels of the network. In the unread pixels, in other words the pixels which are not read during the first and second reading phases, the voltage Vpix is ​​accumulated with the new photo-generated carriers between the period from T2 to T3 . The third read phase involves for example the same voltage ramp from Vsat to Vm as that applied during the first read phase. The integration time T3 is for example chosen to correspond to twice T3 which is a time at which pixels having a read node voltage Vpix just above Vm at time A have a read node voltage Vpix dropping to Vsat at time T\ as represented by line L2. During the third read phase, all pixels having read node voltages between voltages Vsat and Vm will trigger the corresponding comparator and thus be read and deactivated, as for pixels read during the first and second read phases. These pixels correspond to those having read node voltages Vpix between lines L2 and L3 in Figure 6, where L3 represents a pixel having a read node voltage Vpix dropping to Vm at time T3.All other pixels retain for example the stored value Vpix, which for these pixels is between the voltages Vm and Vrst. During the third reading phase, the stamp signal STAMP_INT has for example a third binary value, different from the first and second values ​​and equal for example to binary "01".

[0098] A fourth read phase begins at a time after the TG command has been pulsed again, this TG pulse being for example a global command for all the pixels of the array. In the unread pixels, in other words the pixels which were not read during the first, second and third read phases, the voltage Vpix is ​​accumulated with the new photo-generated carriers between the period from T 3 to E4. The fourth read phase involves for example a full amplitude ramp from Vsat to Vm which is applied in order to read all the pixels not already read during the first, second and third read phases. The integration time Test for example chosen to correspond to twice T3 which is a time at which pixels having a read node voltage just above Vm at time T3 have a read node voltage Vpix dropping to Vsat at time T as represented by line L3.During the fourth reading phase, the stamp signal STAMP_INT has for example a fourth binary value, different from the first, second and third values ​​and equal for example to binary "00".

[0099] An integration period for an (n+l)-th frame (FRAME_n+l) begins for example at time T4.

[0100] [Fig.7] is another timing diagram representing an integration period and reading phases of the image sensor of [Fig.5] according to an exemplary embodiment of the present description. [Fig.7] represents in particular the signal TG, AZ, INIT, V_RAMP, DATA_RAMP, STAMP_INT, WRITE, MEM and DISABLE and a value MEM stored in the memory 108, for a pixel i (PIXEL i) and for a pixel j (PIXEL j).

[0101] In the example of Figure 7, the illumination of pixel i causes it to be read during the first reading phase at time T j and the illumination of pixel j causes it to be read during the fourth reading phase at time T4.

[0102] Initially, before the start of the integration period, the signal DIS AB LE is for example active for all the pixels of the network, the signals TG, AZ, INIT and WRITE, are in the low state and the signal DATA_RAMP is for example at a level corresponding to Vrst, so that the voltage V_RAMP is at the voltage Vrst, or approximately the voltage Vrst.

[0103] At a time t0, the signal STAMP_INT is for example updated to have the value corresponding to the first reading phase, equal to binary "11" in this example. This update occurs for example at any time before the start of the first reading phase.

[0104] At a time tl, the signals AZ and INIT are brought to the high state, causing the logic circuit 110 to deactivate the deactivation signal DISABLE, and causing the read node voltage Vpix to reach the level of the ramp signal V_RAMP which is for example the voltage level Vrst.

[0105] At a time t2, for example after t1, the signals AZ and INIT drop to the low state and the digital ramp signal DATA_RAMP changes for example to a level corresponding to the voltage Vsat, so that the voltage ramp V_RAMP also changes to this voltage Vsat.

[0106] The start time of the first reading phase is for example shortly after time t2 and includes a rising edge of a positive pulse of the TG signal.

[0107] At a time t3 after the time the digital data ramp signal DATA_RAMP begins to increment in binary steps, causing the voltage ramp V_RAMP to increase in steps from Vsat to the voltage Vm. This ramp has for example a reduced amplitude Al compared to a full ramp amplitude.

[0108] At a time t4 after time t3, the voltage level of the voltage ramp V_RAMP crosses the voltage Vpix(i) at the read node SN of pixel i. In this example, in view of the rising voltage ramp, the voltage ramp V_RAMP rises beyond Vpix(i) at time t4, but in the case of a falling ramp, it would also be possible for the voltage ramp V_RAMP to fall below Vpix(i) at time t4. This crossing of the voltage Vpix(i) by the ramp voltage V_RAMP causes the activation of the write signal WRITE_PIXi associated with pixel i and causes the memory 108 of pixel i to store the digital values ​​STAMP_INT + D_PIXi that the stamp signal STAMP_INT and the digital ramp signal DATA_RAMP have at time t4. Furthermore, the deactivation signal DISABLE of pixel i is activated by the logic circuit 110 so that the comparator of pixel i is disabled (COMP PIX i DISABLED) and, in the case of pixel circuit 100, a reset operation is also applied to the photodiode of pixel i.

[0109] At a time t5 after the time t4, the digital ramp signal DATA_RAMP reaches for example a value corresponding to the voltage level Vm and therefore the ramp voltage V_RAMP also reaches the voltage level Vm then drops to return to Vsat, causing the write signal WRITE_PIXi of the pixel i to drop to the low state, thus completing the operation of reading the pixel i in its memory 108.

[0110] The reading of pixel j during the fourth reading phase occurs for example in a similar way to the first reading phase, except that a full voltage ramp is applied.

[0111] the start time 74 of the fourth reading phase comprises a rising edge of a positive pulse of the TG signal.

[0112] At a time t7 after time L4, the digital ramp signal DATA_RAMP begins to increment by binary steps, causing the voltage ramp V_RAMP to increase by steps from Vsat to voltage Vrst. This ramp has for example an amplitude A2 corresponding to the complete range from voltage level Vsat to voltage level Vrst.

[0113] At a time t8 after time t7, the voltage level of the voltage ramp V_RAMP crosses the voltage Vpix(j) at the read node SN of pixel j. This causes the activation of the write signal WRITE_PIXj associated with pixel j and causes the memory 108 of pixel j to store the digital values ​​STAMP_INT + D_PIXj that the stamp signal STAMP_INT and the digital ramp signal DATA_RAMP have at time t8. In addition, the deactivation signal DIS AB LE of pixel j is activated by the logic circuit 110 so that the comparator of pixel j is disabled (COMP PIX j DISABLED) and, in the case of the pixel circuit 100, a reset operation is also applied to the photodiode of pixel j.

[0114] At a time t9 after the time t8, the digital ramp signal DATA_RAMP reaches for example a value corresponding to the voltage level Vrst and therefore the ramp voltage V_RAMP also reaches the voltage level Vrst then drops to return to Vsat, causing the write signal WRITE_PIXj of the pixel j to drop to the low state, thus completing the operation of reading the pixel j in its memory 108.

[0115] As the fourth reading phase is for example the final writing phase of the frame, a pixel reset is then applied for example by activating, at an instant t10, the signals TG and AZ, then by bringing these signals back to the low state at an instant t11. The integration and reading phases of a following frame then begin for example, corresponding to the instant of [Fig.6].

[0116] When an image is to be provided by the image sensor 500, the memory 108 of each of the pixel circuits is for example read, and the digital values ​​are for example reconstructed by the image reconstruction device 516 of [Fig.5], for generate a wide dynamic range image. This reconstruction includes, for example, converting each digital value read from the memory 108 of each pixel circuit into an image pixel value using an appropriate conversion based on the duration of the integration time. The following table (Table 1) provides an example of conversion based on the example of the read phases of Figures 6 and 7: [Tables 1] T DURATIO N ILLUM RANGE MSB ADC BINARY DIG VALUE MIN MAX MIN MAX MIN MAX Tl Tl 1024 2047 11 00000000 11111111 768 1023 T2 2xTl 512 1023 10 00000000 11111111 512 767 T3 4xTl 256 511 01 00000000 11111111 256 511 T4 8xTl 1 255 00 00000000 11111111 1 255

[0117] In this table: T indicates the integration times (T1), 73 (T2), T3 (T3) and (T4); the duration (DURATION) of each integration time is expressed as a multiple of the smallest integration time Tl; the illuminance range (ILLUM RANGE) from minimum (MIN) to maximum (MAX) indicates, in arbitrary units, the relative illuminance range for the image sensor pixels at each period Tl to T4; the two most significant bits (MSB) of the illuminance value, corresponding to the stamp value STAMP_INT, is indicated; the minimum to maximum range of the digital binary value (ADC BINARY) resulting from the ramp conversion is indicated; and the minimum to maximum range of the corresponding digital image pixel value is indicated (DIG VALUE).

[0118] As illustrated in Table 1, the analog to digital conversion is for example arbitrarily set such that, regardless of the ramp amplitude, each full-range maximum signal value corresponds to the maximum binary value and the minimum signal value corresponds to the zero binary value.

[0119] Figure 8 is a graph representing pixel digital values ​​(ADC VALUE (ADU)), similar to the DIG VALUE values ​​of Table 1, in arbitrary digital units, captured by the pixels of the image sensor 500 of Figure 5 as a function of brightness (ILLUMINANCE (AU)) in arbitrary units (AU), similar to the ILLUM RANGE values ​​of Table 1, according to an exemplary embodiment of the present description, and developed on the basis of Table 1 above. This graph represents in particular the compression operation, in particular the representation of a dynamic range of 2047 (about 67 dB), which is eight times that obtained in using a single integration method. Indeed, the longest integration period T4 is eight times longer than the shortest integration period 7].

[0120] Of course, although Figures 6 and 7 are based on four reading phases, a different number of reading phases could be used. An example based on eight reading phases will now be described in connection with [Fig.9].

[0121] [Fig.9] is a graph representing digital pixel values ​​captured by pixels of the image sensor of [Fig.5] as a function of brightness according to another exemplary embodiment of the present disclosure.

[0122] The graph in [Fig.9] is based on a conversion as shown in Table 2 below, which has a similar layout to Table 1, but has eight integration periods T1 to T8, with a factor of four between the duration of one integration period and the next. In particular, period T1 has a nominal duration, period T2 a duration of four times T1, period T3 a duration of four times T2 and therefore 16 times T1, etc. [Tables2] T DURATIO N ILLUM RANGE MSB ADC BINARY DIG VALUE MIN MAX MIN MAX MIN MAX Tl Tl 524288 2097151 111 0 1111111 896 1023 T2 4xTl 131072 524287 110 0 1111111 768 895 T3 16xTl 32768 131071 101 0 1111111 640 767 T4 64xTl 8192 32767 100 0 1111111 512 639 T5 256xTl 2048 8191 11 0 1111111 384 511 T6 1024xTl 512 2047 10 0 1111111 256 383 T7 4096xTl 128 511 1 0 1111111 128 255 T8 16384xTl 0 127 0 0 1111111 0 127

[0123] In this example, the digital value STAMP_INT is a 3-bit value, the conversion length is only 7 bits, and the equivalent dynamic range would reach 127 dB as illustrated by the log-log response in [Fig.9]. The dynamic range improvement in this example has a factor of 16384.

[0124] [Fig. 10] is a graph representing a variable step height of a ramp signal according to an exemplary embodiment of the present disclosure. [Fig. 10] represents in particular the step height k (STEP HEIGHT k) as a function of the number of steps (STEP NO.). Such a variation of the step height ensures an improvement in the conversion speed, by decreasing the ramp time. This is based on the fact that high illumination levels contain a strong shot noise component and therefore there is no gain in converting the full dynamic range with fine resolution.

[0125] For example, considering a maximum electronic capacitance of the readout node of a pixel circuit of 6000e-, the corresponding shot noise is ^6000 = 77.5 e' . In a 10-bit conversion scale, this is equivalent to 13.2 ADU. This implies that at least 3 bits of the converted signal represent noise. At half the dynamic range, i.e., at 3000e-, the shot noise is 9.3 ADU.

[0126] An improved ramp staging is for example described in the publications of Snoeji MF, et al. “A low power column parallel 12-bit ADC for CMOS imagers”, in IEEE workshop on CCDs 8 AIS, 2005, pl69-172, and Snoeji MF, et al. “Multiple-ramp column-parallel ADC architectures for CMOS image sensors” in IEEE journal of solid-state image circuits, vol. 42, no. 12, pp 2968-2977. Based on the following equation cited in the latter publication:

[0127] [Math.l] AT __ / ] \ 2".^2.r )

[0128] with Nsat = 6000e-, n = 10 bits, r= 0.1 and k=(l,2,4,8), only x2 and x4 steps are useful for converting half-dynamic cycles. Calculations indicate that less than 200 steps are sufficient for a compatible 10-bit resolution for the first three cycles. The same configuration can be used for the last conversion, where a full range is scanned.

[0129] Calculations lead to a total of 301 variable steps distributed as shown in [Fig. 10]. In particular, in this example, the conversion starts with 31 steps at eight times the nominal step height, followed by 147 steps at eight times the nominal step height, then 49 steps at the nominal step height, leading to a total of 301 steps instead of 1024 steps. Such an implementation is for example carried out using a variable step DAC controlled by a programmed counter.

[0130] Second aspect - Pixel circuit with event-driven reading

[0131] Although the embodiments described in connection with Figures 1 to 10 above are based on the use of a memory 108 implemented in each pixel circuit of an image sensor, in some embodiments there is not enough space for such a memory. Furthermore, the power consumption of the parallel data bus used to transmit the digital ramp signal DATA_RAMP to all pixels in the array in parallel may be too high for some applications, given the relatively high capabilities to be driven at each node of the data bus. An event-driven solution will now be described in connection with Figures 11 to 20 which provide an alternative approach devoid of the memory 108 and the digital ramp signal DATA_RAMP.

[0132] [Fig. 11] schematically represents a pixel circuit 1100 based on event-driven reading according to an exemplary embodiment of the present disclosure. Pixel 1100 is similar to pixel 100 of [Fig. 1] and like elements are designated by like references and will not be described herein again in detail. In particular, the sensor 102, the comparator 104 and the automatic zeroing switch 112 are the same as those of the pixel 100. However, the logic circuit 110 and the memory 108 of the pixel circuit 100 are replaced in the pixel circuit 1100 by an event signaling circuit (EVENT SIGNALING) 1102, and the output signal of the comparator 104 is no longer a write signal, but an event signal EVENT indicating that an event should be communicated to a peripheral event recording circuit (described in more detail later).Indeed, the event signaling circuit 1102 implements for example an authentication protocol on one or more row lines (ROW SIGNALING) which are used for signaling the row line and on one or more column lines (COLUMN SIGNALING) which are used for signaling the column line.

[0133] The event signaling circuit 1102 also receives, for example, the initialization signal INIT and, in response to this command, is configured, for example, to restore the functionality of the comparator 104 and / or the photodiode PD by placing a deactivation signal DISABLE in a non-activated state. The deactivation signal DIS AB LE is, for example, provided to an activation input of the comparator 104 and / or is used to activate the reset transistor MRST, if present.

[0134] [Fig. 12] schematically represents the pixel circuit 1100 of [Fig. 11] in more detail according to another exemplary embodiment of the present disclosure, based on three stages TIER 1, TIER 2 and TIER 3. The pixel circuit of [Fig. 12] is similar to the pixel circuit of [Fig.2], and the TIER 1 and TIER 2 stages of [Fig. 12] comprise the same elements as those of [Fig.2]. The TIER 3 stage of [Fig. 12] is, however, different from that of [Fig.2], the output of the Schmitt trigger 206 generating the event signal EVENT and being connected to the event signaling circuit 1102.

[0135] An exemplary embodiment of the event signaling circuit 1102 is illustrated in [Fig. 12]. For example, the event signaling circuit 1102 is similar to that described in the US patent application published under number US2021 / 0337150 in the name of the present applicant.

[0136] The one or more column and row signal lines respectively comprise, in the embodiment of [Fig. 12], shared column lines 1204 and 1206 and shared row lines 1208 and 1210. The shared column lines 1204, 1206 are for example connected to at least two of, and generally to all of, the pixels of a column of the pixel array of the image sensor and the shared row lines 1208, 1210 are for example connected to at least two of, and generally to all of, the pixels of a row of the pixel array of the image sensor.

[0137] The shared column line 1204 is for example a column read request line on which the event signaling circuit 1102 indicates when it detects an event by activating a column request signal. The shared column line 1206 is for example an acknowledgement line on which the event signaling circuit 1102 receives a column acknowledgement signal ackx indicating that the request has been registered by a corresponding column event detection circuit (described in more detail later).

[0138] The shared row line 1208 is, for example, a row read request line on which the event signaling circuit 1102 indicates when it detects an event and has received an acknowledgment ackx on the shared column line 1206, by activating a column request signal r('Qy. The shared row line 1210 is, for example, an acknowledgment line on which the event signaling circuit 1102 receives a row acknowledgment signal acky indicating that the request has been registered by a corresponding row event detection circuit (described in more detail later).

[0139] The event signaling circuit 1102 is for example configured, for example, in response to the activation of the event signal EVENT by the comparator, the request signal recix on the line 1204. The event signaling circuit 1102 is then, for example, configured to detect when the acknowledgment signal ackx is activated on the line 1206, and to activate, in response, the request signal reSy on the line 1208. The event signaling circuit 1102 is then, for example, configured to detect when the acknowledgment signal ackx is activated on the line 1210, and to activate, in response, the deactivation signal, which in the example of FIG. 12 is a DIS ART signal which is active in the low state.

[0140] The output of the comparator, corresponding for example to the output of the Schmitt flip-flop 206, is for example connected to an input of a flip-flop 1214. The flip-flop 1214 is configured to store an event detection state at its output Q until the event has been recorded by the row and line event detection circuits. In certain embodiments, the flip-flop 1214 is implemented by a reset-set (RS) flip-flop in which the set input S of the flip-flop 1214 receives the output of the Schmitt flip-flop 206. However, other implementations based on any bistable device would be possible.

[0141] In some embodiments, the output of flip-flop 1214 is connected to one input of an AND gate 1216, the other input of which is connected to the output of an additional flip-flop 1218. The additional flip-flop 1218 receives, for example, at one input, the INIT signal, indicating when the pixel circuit 1200 is to become active. In some embodiments, flip-flop 1218 is implemented by an RS flip-flop 1218 receiving the INIT signal at its set input.

[0142] The output of AND gate 1216 activates, for example, a switch 1220 which is configured to activate the request signal reqx. In some embodiments, the request signal reQx is active in the low state, and the switch 1220 is implemented by a transistor, such as an NMOS transistor, having its gate connected to the output of AND gate 1216 and connecting, via its main conduction nodes, line 1204 to ground. Of course, it would also be possible for line 1204 to be active in the high state, by replacing NMOS transistors with PMOS transistors and inverting their respective digital input control.

[0143] The output of AND gate 1216 also causes the request signal reLiy on line 1208 to be activated when the acknowledgement signal ackx on line 1206 is activated. For example, the request signal re <iy est activé lorsque deux commutateurs 1224, 1226 sont activés, le commutateur 1224 étant activé par la sortie de la porte ET 1216 et le commutateur 1226 étant activé par le signal d'accusé de réception ackx. Dans certains modes de réalisation, le signal de requête reQy est actif à l'état bas et les commutateurs 1224, 1226 sont reliés en série l'un à l'autre entre la ligne 1208 et la masse. Par exemple, les commutateurs 1224, 1226 sont mis en œuvre par des transistors, tels que des transistors NMOS, reliés en série par l'intermédiaire de leurs nœuds principaux de conduction, le commutateur 1224 ayant sa grille reliée à la sortie de la porte ET 1216, et le commutateur 1226 ayant sa grille reliée à la ligne 1206.Of course, it would also be possible to make line 1208 active high, by replacing the NMOS transistors with PMOS transistors and inverting their respective digital input control.

[0144] The resetting of the event signaling circuit 1102 is for example carried out by an AND gate 1230 which has for example one of its inputs connected to the acknowledgment line 1206, the other of its inputs connected to the acknowledgment line 1210 and its output connected to the flip-flops 1214, 1218 in order to reset their states. In the case where the flip-flops 1214, 1218 are RS flip-flops, the output of the AND gate 1230 is for example connected to the reset input R of these flip-flops.

[0145] [Fig. 13] schematically represents a pixel circuit 1300 according to yet another exemplary embodiment of the present disclosure. The pixel circuit 1300 is similar to pixel circuit 1200, but with the same modifications as those applied in pixel circuit 400 of [Fig.4] compared to pixel circuit 200 of [Fig.2]. Pixel circuit 1200 could also be implemented according to the intermediate solution of [Fig.3].

[0146] [Fig. 14] schematically represents an image sensor 1400 comprising an array of pixel circuits according to another exemplary embodiment of the present disclosure. The pixel circuits are for example implemented by the pixel circuit 1100 of [Fig. 11] or 12 or by the pixel circuit 1300 of [Fig. 13]. In particular, the image sensor 1400 comprises for example a sensor array (SENSOR ARRAY) 1402 comprising, for example, one or more stages which comprise the sensors 102 and at least a part of the comparators of the pixel circuits and a logic array (LOGIC ARRAY) 1404 stacked on the sensor array 1402 and comprising, for example, the event signaling circuit 1102 of the pixel circuits and in some cases a part of the comparators 104 of the pixel circuits.

[0147] The logic network 1404 comprises, for example, a peripheral logic circuit for registering an event. In particular, a column event detection circuit (X REGISTER) 1406 is, for example, connected to the column signaling lines, such as column lines 1204, 1206, of each column of pixel circuits, and a row event detection circuit (Y REGISTER) 1408 is, for example, connected to the row signaling lines, such as row lines 1208, 1210, of each row of pixel circuits. The column and row event detection circuits 1406, 1408 are, for example, configured to process events signaled by pixel circuits of the array one by one as they occur and to provide arbitration between the events so that all events are correctly processed. For example, the implementation described in US2021 / 0337150 mentioned above may be used.

[0148] A column encoder (X ENCODER) 1410 is for example connected to the column event detection circuit 1406 and generates a column address X ADDRESS. A row encoder (Y ENCODER) 1412 is for example connected to the row event detection circuit 1408 and generates a row address Y ADDRESS.

[0149] The image sensor 1400 also comprises, for example, a counter (COUNTER) 1414, which is, for example, an 8-bit, 10-bit or 12-bit counter, configured to generate the digital ramp signal DATA_RAMP, which is not supplied to the pixel circuits of the arrays 1402, 1404, but to a DAC 1416 and to an image memory 1418. The DAC 1416 is configured to convert the digital ramp signal DATA_RAMP into a voltage signal in order to generate the voltage ramp V_RAMP supplied to the sensor array 1402 and, in particular, to the comparator 104 of each of the pixel circuits. The image memory 1418 is for example a memory integrated in the image sensor 1400 or could alternatively be an external memory, i.e. a memory which is external to the image sensor 1400.

[0150] The image sensor 1400 also comprises, for example, a control circuit (CONTROL) 1420, implemented, for example, by a finite state machine and configured to generate the signal INIT, which is supplied to the logic network 1404 and, in particular, to the event signaling circuit 1102 of each pixel circuit and the signals TG, AZ and V_BIAS which are supplied to the sensor network 1402 and, in particular, to the sensor 102, the automatic zeroing switch 112 and the comparator 104 of each pixel circuit. The control circuit 1420 also generates, for example, the signal STAMP_INT, which is supplied to the image memory 1418 and a clock signal CK and a digital signal D_VRST to the counter 1414. The digital signal D_VRST sets an initial value of the counter 1414 and the clock signal CK controls when the counter 1414 increments its count.The control circuit 1420 is for example configured to receive an end of scan signal (END_OF_SCAN) from the column or row event detection circuit 1406 or 1408 indicating when all pixel circuits signaling an event have been processed.

[0151] The digital signals DATA_RAMP and STAMP_INT together constitute a digital value PIXVAL which is stored in the image memory 1418 under the control of a write signal WRITE generated by the control circuit 1420. The value PIXVAL is written to an address in the image memory 1418 indicated by the column and row addresses X ADDRESS, Y ADDRESS. For example, the image memory 1418 includes an address corresponding to each pixel circuit in the array.

[0152] Although not illustrated in [Fig. 14], the image sensor 1400 for example also includes an interface for reading the memory 1418 and / or an image reconstruction device for reconstructing wide dynamic range images, similar to elements 514 and 516 of [Fig. 5].

[0153] The operation of the image sensor 1400 will now be described in more detail in relation to the timing diagram of [Fig. 15].

[0154] [Fig. 15] is a timing diagram representing reading phases of the image sensor of [Fig. 14] according to an exemplary embodiment of the present description. [Fig. 15] represents in particular the TG signal, the INIT and AZ signals which are for example the same signal in event-driven solutions, the V_RAMP signal, the DATA_RAMP signal, the STAMP_INT signal and the X ADDRESS and Y ADDRESS signals.

[0155] Before the first read cycle, the voltage ramp V_RAMP is for example brought from the voltage level Vsat to the voltage level Vrst at a time t0. The signals INIT and AZ are then pulsed at a time tl while VRAMP = Vrst to initialize the pixel. The signal TG is then pulsed at the start time T j of a first reading phase and the reading node voltage Vpix is ​​generated at the comparator input.

[0156] At a time t2, the digital signal STAMP_INT is for example set to a value associated with the first reading phase.

[0157] At a time t3, the digital ramp DATA_RAMP and the voltage ramp V_RAMP begin to increment in steps. For each step, all pixel circuits having their read node voltage Vpix which is crossed by the voltage ramp V_RAMP generate an event. All corresponding addresses of these pixel circuits are coded sequentially or in parallel, depending on the architecture considered, by individually terminating the authentication protocol. These addresses, as well as the current digital values ​​DATA_RAMP and STAMP-INT, are read and / or stored in the image memory 1418, under the control of the write signal (WRITE) transmitted by the control circuit 1420 to the memory 1418. For example, the write signal WRITE is activated in response to each significant edge of a clock signal clk while the end of scan signal END_OF_SCAN is not activated.The clock signal clk is for example the clock which is used to time synchronize the column and row event detection circuits 1406, 1408 and thus indicates the read speed of the pixel array.

[0158] Once there are no more pixel requests to be processed, the end of scan signal END_OF_SCAN is for example activated and the control circuit 1420 is configured to increment the digital ramp signal DATA_RAMP by one step causing the voltage ramp V_RAMP to increase by a voltage level equivalent to one LSB. The cycle then begins again. The amplitudes of the digital ramp signal DATA_RAMP and of the voltage ramp V_RAMP during the following reading phases are for example chosen in the same way as described previously, all the ramps rising for example to an intermediate value Vm, except the ramp of the final reading phase of the frame, which is for example a full range ramp.

[0159] It will be noted that the variable duration of each step of the voltage ramp V_RAMP could lead, in the case of relatively high illumination levels in the image, to a situation in which the first reading phase is not completed by the instant T 2 at which the second reading phase must begin. Certain techniques for overcoming this situation will now be described in relation to Figures 16 to 20.

[0160] Figure 16 is another timing diagram representing reading phases of the image sensor of Figure 14 according to an exemplary embodiment of the present disclosure. Figure 16 is similar to the diagram of Figure 6, except that the read phases of a single frame are illustrated in Figure 16 and the time synchronization of the second read phase is modified. In particular, if the first read phase extends beyond the start time 73 of the second read phase, the second read phase is delayed to begin at a time T2.

[0161] Of course, changing the start of the second read phase involves changing the compression curve used to convert the read node voltage Vpix into the image pixel value. In addition, there is a risk that some pixel circuits may saturate, causing a loss of information.

[0162] Figure 17 is another timing diagram showing readout phases of the image sensor of Figure 14 according to another exemplary embodiment of the present disclosure. Figure 17 is similar to the timing diagram of Figure 16, except that, rather than delaying the second readout phase, the first readout phase is divided into two parts a and b at times T\a and T\b respectively. In this case, only the AZ signal is generated at times T\a and T\b and no INIT is used, the pixels having been read remaining disabled. This approach may work well in some applications, but for some applications it could lead to the afterimage effect for very fast moving scenes, due to the reinitialization by the AZ signal at two different times.

[0163] Figure 18 is another timing diagram showing readout phases of the image sensor of Figure 14 according to yet another exemplary embodiment of the present disclosure. The timing diagram of Figure 18 is similar to that of Figure 6 and shows pixel values ​​(PIXEL VALUES) based on readout node voltages over two consecutive frames n (FRAME_n) and n+1 (FRAME_n+1). However, in the example of Figure 18, the readout phases at times T1 and T2 involve sampling the readout node voltage in each pixel circuit and storing the presence of events locally, without reading the event using the event signaling circuit. The readout is performed after the third readout phase (READ PHASE 3), as there is a larger time interval between times T3 and T4 in which to perform this operation.The final reading phase (READ PHASE 4) takes place, for example, as in the previous examples.

[0164] According to the solution of Figure 19, the input voltage ramp of the comparator V_RAMP is for example fixed at a constant level of Vm during the first and second phases and pixels having values ​​between this value and Vsat at the end of each integration time T2 will be stamped with the corresponding STAMP_INT value and the corresponding TG command is for example deactivated until the end of the frame cycle. Therefore, the read node voltage Vpix is ​​stored at the read node SN of these pixels until the third read phase at time T3 , at which time all pixels having events at times T15 T 2 or T3 are read. At time T the remaining pixels can be read with the full dynamic range applied to the voltage ramp V_RAMP, as explained previously. At this time, or optionally once the final read phase is completed, the integration phase of a new frame begins for example.

[0165] The local storage in the pixel circuits of the stamp value STAMP_INT is for example carried out using an internal register and a logic circuit is for example also added in each pixel in order to deactivate the arrival of the TG pulse for pixels detecting events during the first and second reading phases as will now be described in relation to [Fig. 19].

[0166] Figure 19 schematically represents a pixel circuit 1900 according to yet another exemplary embodiment of the present description. The pixel circuit 1900 has many elements common to the circuit 1100 of Figure 12 and these elements are designated by the same references and will not be described again in detail. However, in the pixel circuit 1900, the comparator output, corresponding for example to the output of the Schmitt trigger 206, is switched by a flip-flop 1902, the output of which deactivates the input command TG arriving at the pixel. For example, the flip-flop 1902 is an RS flip-flop and an inverter output Q of this flip-flop is for example connected to an input of an AND gate 1904, the other input of which is connected to the line TG receiving the signal TG.The output of AND gate 1904 is connected, for example via interstage connections 212, 210 (instead of the disable signal), to the control output of transfer gate 106. Therefore, if the voltage at read node Vpix reaches Vm during the first and second read phases, transfer gate 106 will remain disabled until a subsequent INIT command. A downward arrow in Schmitt trigger 206 and a rising arrow in AND gate 1904 indicate a level shift up or down in order to accommodate the two different power supplies.For example, supply and signal levels in the TIER 3 stage are relatively low voltages, such as 1 V or 1.2 V, since this stage is for example implemented in an advanced logic process, whereas in the TIER 1 and TIER 2 stages, devices are for example implemented in accordance with an analog process with a higher supply voltage, such as 2.5 V.

[0167] An OR gate 1906 is also for example added in the pixel circuit 1900 with respect to the pixel circuit 1200, this OR gate having one input receiving an input signal INH from the pixel circuit, and its other input connected to the output of the AND gate 1230. This input signal INH which is generated for example by the control circuit 1420 is for example used to disable the event signaling circuit 1102 until the reading is to be performed during the third reading phase at time 7\

[0168] A 3-bit register (3-BIT REG) 1908 is also added, this register receiving, at a data input, the stamp signal STAMP_INT and being controlled by the non-inverting output Q of the flip-flop 1902 to store the value of the signal STAMP_INT when an event is detected during the first or second reading phase. An output of the register 1908 is connected to a data bus STAMP_BUS which is for example common to the pixel circuits of the same row of the array. The register 1908 is for example controlled to provide as output its stored value by an output of an AND gate 1910, which has one of its inputs connected to the output of the AND gate 1216 and its other input connected to the line 1206 receiving the acknowledgment signal ackx. For example, the STAMP_BUS data bus is a high impedance bus that can for example be implemented using transistors similar to transistor 1220 to lower the voltage on the bus.

[0169] Either of the readout synchronization solutions described in connection with Figures 16 to 19 may be preferable depending on the number of pixel circuits that have high or low illumination values. In some event-driven architecture solutions, a real-time image histogram is for example generated concurrently with the image frame capture operation and used for example to adjust the exposure time ratios and / or the readout mode for subsequent frames, as will now be described in more detail in connection with [Fig. 20].

[0170] [Fig. 20] is a timing diagram showing readout phases of the image sensor of [Fig. 14] according to yet another exemplary embodiment of the present disclosure. The timing diagram of [Fig. 20] is identical to that of [Fig. 15], except that a PIXEL COUNTER signal is added, representing the generation of the image histogram. For example, at each step of the digital ramp signal DATA_RAMP during each of the readout phases, an event counter is used to count the number of pixel circuits having events occurring at the given level and, when the digital data ramp signal DATA_RAMP is incremented, the count value is, for example, stored in a memory (not shown). For example, this event counter is part of the control circuit 1420 (represented by a dotted block EVENT COUNTER 1422 in [Fig.14]) and is incremented each time the WRITE signal is activated, indicating that an event has been detected. At the end of the frame cycle, the count values ​​reached by each counter for the different levels of the digital ramp signal provide the image histogram (HISTOGRAM), of which . An example is shown at the bottom of [Fig.20]. This pixel value distribution graph can be used to calculate the integration and conversion parameters applied to one or more subsequent image frames. For example, a dedicated processor is used to calculate the next frame configuration. Common aspects

[0171] An advantage of the pixel circuits and image sensors described herein is that wide dynamic range images can be generated without significantly increasing the pixel footprint and with relatively reduced power consumption. Furthermore, post-production processing can be performed to reconstruct the wide dynamic range images based on the read pixel values.

[0172] An advantage of performing the analog to digital conversion locally in each pixel and storing the digital pixel values ​​in local memory as described in connection with Figures 1-10 is that the pixels are almost immediately read out after a charge transfer. This feature avoids the need for restricted specifications to account for, for example, leakage at the readout node as well as stray light sensitivity (PLS), which can be critical in a conventional snapshot (global shutter) operation, where the readout node is left floating for an entire frame length that is typically between 10 ms and 20 ms. Since the readout node is a high impedance mode, it is also very sensitive to glitches and noise.

[0173] An advantage of resetting and deactivating the photodiodes of pixels that have been read is that it avoids saturation of the photodiodes and the risk of charges from saturated photodiodes migrating to surrounding pixels.

[0174] Various embodiments and variations have been described. Those skilled in the art will understand that certain features of these various embodiments and variations could be combined, and other variations will be apparent to those skilled in the art. For example, although embodiments are described in which a ramp voltage V_RAMP is used to perform analog-to-digital conversion or event detection, an alternative threshold detection mode could also be implemented in which a constant voltage, e.g., at voltage level Vm, is applied instead of the voltage ramp, so that pixels that are illuminated beyond a corresponding threshold can be detected. This could, for example, be used in the detection of bright objects such as vehicle headlights, or other traffic lights in the image scene.In the event-driven configuration, the added value is that a direct pixel address detection that indicates the object position can be achieved, without reading the entire memory.

[0175] Further, although some examples of reduced ramp amplitudes have been provided, there are many possible values. For example, according to one mode of rea- lization, a first ramp amplitude could be 25 percent of the full range and a second amplitude 50 percent.

[0176] Furthermore, it will be apparent to those skilled in the art that the various circuits could be modified to replace the NMOS transistors with PMOS transistors or vice versa and different transistor technologies could be used.

[0177] Finally, the practical implementation of the embodiments and variants described is within the reach of those skilled in the art from the functional indications given above.

Claims

Claims

1. Pixel circuit of an image sensor, the pixel circuit comprising: a pinched photodiode (PD) connected to a read node (SN); a comparator (104) configured: - to compare, during a first read phase, a first read node voltage (Vpix) at the read node (SN), resulting from a first integration period of the pinched photodiode (PD), with a first voltage ramp (V_RAMP) and to generate an event signal (EVENT) if the voltage of the first voltage ramp (V_RAMP) crosses the first read node voltage (Vpix);and, - if the voltage of the first voltage ramp (V_RAMP) does not cross the first read node voltage (Vpix) during the first read phase, to compare, during a second read phase, a second read node voltage (Vpix) at the read node (SN), resulting from the first integration period and a second integration period of the pinched photodiode (PD), with a second voltage ramp (V_RAMP) and to generate an event signal (EVENT) when the voltage of the second voltage ramp (V_RAMP) crosses the second read node voltage (Vpix);and an event signaling circuit (1102) configured to signal the generation of the event signal during the first or second read phase to an event recording circuit, wherein a cathode of the photodiode (PD) is connected to the read node (SN) via a transfer gate (106), the transfer gate being activated by a first activation pulse at the start of the first read phase and by a second activation pulse at the start of the second read phase, the voltage at the read node (SN) not being reset between the first and second read phases.;

2. The pixel circuit of claim 1, wherein the event signaling circuit (1102) is further configured to disable the comparator (104) during the second read phase if the event signal (EVENT) is generated during the first read phase.

3. A pixel circuit according to claim 1 or 2, wherein the second ramp signal has an amplitude (A2) greater than the amplitude (Al) of the first ramp signal.

4. A pixel circuit according to any one of claims 1 to 3, further comprising a reset transistor (MRST) connecting the cathode of the photodiode (PD) to a reset voltage rail (Vrst), wherein the event signaling circuit (1102) is further configured to activate the reset transistor (MRST) in response to the event signal (EVENT) generated by the comparator (104) during the first or second read phase.

5. A pixel circuit according to any one of claims 1 to 4, further comprising a logic device (1904) connected to the output of the comparator (104) and configured to deactivate the transfer gate (106) in response to the generation of the event signal (EVENT) during the first or second read phase.

6. An image sensor comprising: - an array consisting of a plurality of the pixel circuit according to any one of claims 1 to 5; - a counter (1414) configured to generate a digital ramp (DATA_RAMP) and to provide the digital ramp to an image memory (1418); and - a digital to analog converter (1416) configured to convert the digital ramp (DATA_RAMP) into the first voltage ramp (V_RAMP) during the first reading phase and into the second voltage ramp during the second reading phase.

7. The image sensor of claim 6, further comprising an image reconstruction device (516) configured to reconstruct an image by generating: - a first pixel value of the image based on a first value of the digital ramp associated with a first circuit of the plurality of pixel circuits; and - a second pixel value of the image based on a second value of the digital ramp associated with a second circuit of the plurality of pixel circuits and based on the relative durations of the first and second integration periods.

8. An image sensor according to claim 6 or 7, further comprising a control circuit (1420) configured, during the first and second read phases, to interrupt the incrementation of the counter (1414) at each count value of the digital ramp until a corresponding end-of-scan signal (END_OF_SCAN) is set by the event recording circuit indicating that all pixel circuits generating the event signal (EVENT) corresponding to the count value have been read.

9. The image sensor of claim 8, wherein the control circuit (1420) is configured to delay a start of the second read phase until all pixel circuits having events have been read during the first read phase.

10. Image sensor according to claim 8, wherein the control circuit (1420) is configured to perform the first reading phase in two parts, a first part after a first integration time (Tal) and a second part after a second integration time (T„2).

11. An image sensor according to any one of claims 8 to 10, wherein the control circuit (1420) is configured to generate an image histogram by counting the number of pixel circuits signaling events during each count value of the counter (1414).

12. An image sensor according to any one of claims 6 to 11, the image memory (1418) is further configured to receive a digital signal (STAMP_INT) having a first value during the first reading phase and a second value during the second reading phase, the image memory (1418) being configured to store, in response to the event signal (EVENT) from one of the pixel circuits during the first or second reading phase, the value of the digital signal (STAMP_INT).

13. An image sensor according to any one of claims 6 to 12, wherein each pixel circuit further comprises a register (1908) connected to the output of the comparator (104) and configured to store, in response to the generation of the event signal (EVENT), a digital signal (STAMP_INT) having a first value during the first reading phase and a second value during the second reading phase.

14. An image sensor according to any one of claims 6 to 13, comprising first and second stacked stages (TIER 1, TIER 2), each of the pixel circuits of the array comprising a sensor (102), the sensor (102) comprising the pinched photodiode (PD) and the readout node (SN) of each pixel circuit, the sensor (102) being formed in the first stage (TIER 1) and a portion of the comparator being formed in the second stage (TIER 2).

15. An image sensor according to claim 14, wherein the si- circuit Event signaling (1102) is formed in the second stage (TIER 2), or in a third stage (TIER 3) of the image sensor stacked with the first and second stages (TIER 1, TIER 2).

16. A method of reading a pixel circuit of an image sensor, the pixel circuit comprising a pinched photodiode (PD) connected to a reading node (SN), the method comprising: - comparing, by a comparator (104) during a first reading phase, a first reading node voltage (Vpix) at the reading node (SN), resulting from a first integration period of the pinched photodiode (PD), with a first voltage ramp (V_RAMP) and generating, by the comparator (104), an event signal (EVENT) if the voltage of the first voltage ramp (V_RAMP) crosses the first reading node voltage (Vpix); - if the voltage of the first voltage ramp (V_RAMP) does not cross the first read node voltage (Vpix) during the first read phase, comparing, by the comparator (104) during a second read phase, a second read node voltage (Vpix) at the read node (SN), resulting from the first integration period and a second integration period of the pinched photodiode (PD), with a second voltage ramp (V_RAMP) and generating an event signal (EVENT) when the voltage of the second voltage ramp (V_RAMP) crosses the second read node voltage (Vpix); and signaling, by an event signaling circuit (1102), the generation of the event signal during the first or second reading phase to an event recording circuit, wherein a cathode of the photodiode (PD) is connected to the reading node (SN) via a transfer gate (106), the transfer gate being activated by a first activation pulse at the start of the first reading phase and by a second activation pulse at the start of the second reading phase, the voltage at the reading node (SN) not being reset between the first and second reading phases.

17. The method of claim 16, further comprising: -receiving, by an image memory (1418) of the image sensor, a digital ramp (DATA_RAMP); and - storage, at an address of the image memory (1418) associated with the pixel circuit and in response to the event signal (EVENT) of the com parator (104) during the first or second reading phase, of a value of the digital ramp (DATA_RAMP) to constitute pixel data (DATA_PIX) of the pixel.