Method and device for detecting the temporal variation of the light intensity in a matrix of photosensors

Abstract

The invention relates to a method and a device for detecting the temporal variation of the light intensity in a matrix of photosensors, comprising a matrix of pixels, a block for the automatic adjustment of the amplification of the photocurrent, and an arbitrating and event-encoding block. Each pixel comprises a photosensor that generates a photocurrent, an adjustable gain current mirror connected to the outlet of the photosensor, a transimpedance amplifier arranged at the outlet of the current mirror, optionally at least one amplification circuit arranged at the outlet of the transimpedance amplifier, and capacitors and detectors of thresholds for determining whether the output voltage exceeds a higher threshold or drops below a lower threshold in order to generate an event in the pixel.

Claims

1. A device for detecting temporal variation of the light intensity in a matrix of photosensors, characterised in that it comprises at least: a matrix of pixels, wherein each pixel comprises at least: a) a photosensor which generates a photocurrent proportional to a light striking its surface; b) an adjustable gain current mirror (8) comprising an input branch, a first output branch with adjustable current gain and a second output branch with a fixed gain, which copy the input photocurrent to respective outputs and where the output branch with fixed gain is connected to a collector transistor (T4c.sub.1) connected in diode configuration and whose nodes are connected to the collector transistors of the other pixels of the matrix; c) a transimpedance amplifier (T6a.sub.1-T6d.sub.1) arranged in the output of the adjustable gain current mirror, the amplifier comprising at least two MOS transistors polarised in weak inversion and arranged in series, each MOS transistor being in diode configuration, for the conversion of the photocurrent into a logarithmic voltage; d) a switched capacitor circuit (14) comprising a first capacitor (C4) connected to the output of the transimpedance amplifier (T6a.sub.1-T6d.sub.1), a voltage amplifier (T10a.sub.1-T10b.sub.1) connected to the first capacitor (C4), and a second capacitor (C3) connected in series to the first capacitor (C4) and feedback connected to the voltage amplifier, the second capacitor being connected in parallel to a MOS transistor (T11.sub.1) which acts as a reset key; and, e) a first threshold detector to determine if the voltage exceeds a higher threshold and a second threshold detector to determine if the voltage drops below a lower threshold, both detectors being connected to the output of the second voltage amplifier (T10a.sub.1-T10b.sub.1) and said higher and lower thresholds being previously set by a user, a block for the automatic adjustment of the amplification of the photocurrent, which calculates the average of the photocurrents of the pixels of the matrix; and, an arbitrating and event-encoding block connected to the output of the pixel matrix.

2. The device for detecting temporal variation of the light intensity according to claim 1, characterised in that the transimpedance amplifier (T6a.sub.1-T6d.sub.1) is connected to switched capacitor circuit (14) by interposing at least one additional amplification block (10), the amplification blocks (10) being connected in cascade or iteration, wherein the input of the first additional amplification block (10) is connected to the output of the first transimpedance amplifier (T6a.sub.1-T6d.sub.1) and the output of the last additional amplification block (10) is connected to the first capacitor (C4) of the switched capacitor circuit (14), wherein each block comprises at least one transconductance amplifier (11), a fixed gain current mirror (12) connected to the output of the transconductance amplifier (11) and an additional transimpedance amplifier (T9a.sub.1-T9c.sub.1) with at least two additional MOS transistors polarised in weak inversion and connected in diode configuration, the second transimpedance amplifier being connected to the output of the fixed gain current mirror.

3. The device for detecting temporal variation of the light intensity, according to claim 2, characterised in that when there is more than one additional amplification block (10), the blocks (10) are connected in cascade or iteration one to the other, by connecting the gate terminal of the transconductance amplifier (11) of each block with the output of the additional transimpedance amplifier (T9a.sub.1-T9c.sub.1) of the previous block.

4. The device for detecting temporal variation of the light intensity, according to claim 3, characterised in that the adjustable gain current mirror is formed by at least one MOS input transistor (T4a.sub.1), one MOS output transistor (T5.sub.1) and a voltage inversion amplifier (T1.sub.1-T3.sub.1).

5. The device for detecting temporal variation of the light intensity, according to claim 2, characterised in that the adjustable gain current mirror is formed by at least one MOS input transistor (T4a.sub.1), one MOS output transistor (T5.sub.1) and a voltage inversion amplifier (T1.sub.1-T3.sub.1).

6. A method for detecting the temporal variation of the light intensity in a matrix of photosensors, which uses the device described in claim 2, characterised in that in each pixel of the matrix, the following stages are carried out: 1) transforming the light striking the pixel in a current I.sub.ph by means of the photodiode; 2) amplifying the current I.sub.ph to a value A.sub.II.sub.ph by means of the adjustable gain current mirror; 3) adapting the value A.sub.I so that the average A.sub.II.sub.ph remains constant against the temporal variations of the average brightness of all the pixels by means of the automatic amplification block of the photocurrent, which adjusts the amplification in stage 2; 4) converting the current A.sub.II.sub.ph adapted to a voltage by means of the transimpedance amplifier (T6a.sub.1-T6d.sub.1), which comprises a plurality of MOS transistors polarised in weak inversion and connected in series, each of them being connected in diode configuration; 5) determining in the switched capacitor circuit a voltage difference V=V(t.sub.2)V(t.sub.1) between two consecutive times t.sub.1 and t.sub.2, caused by a temporal variation of the light intensity and comparing the voltage difference with a fixed positive reference value V.sub.R+ and a fixed negative reference value V.sub.R, being V.sub.R+ and V.sub.R the same for all pixels of the matrix; 6) generating the digital signal s which is sent to the arbitrating and event-encoding block, the signal being selected among: a positive event every time the first threshold detector determines that the voltage exceeds the higher threshold, generated in the first threshold detector; and, a negative event every time the second threshold detector determines that the voltage drops below the lower threshold, generated in the second threshold detector; and; in the arbitrating and event-encoding block connected to the output of the pixel matrix the following stages are carried out: identifying the spatial coordinates (x,y) of the pixels of the matrix which have generated a digital signal, sending to an external device an event containing the spatial coordinates (x,y) and the signal s; and, generating a flow of events (x,y,s) representing the temporal variation of the light intensity on the matrix of photosensors.

7. The device for detecting temporal variation of the light intensity, according to claim 1, characterised in that the adjustable gain current mirror is formed by at least one MOS input transistor (T4a.sub.1), one MOS output transistor (T5.sub.1) and a voltage inversion amplifier (T1.sub.1-T3.sub.1).

8. The device for detecting temporal variation of the light intensity according to claim 7, characterised in that in the MOS input transistor (T4a.sub.1) of the adjustable gain current mirror has: its gate terminal connected to a voltage VG previously set by a user from outside the device; its drain terminal connected to the photosensor; and, its source terminal connected to the output of the voltage inversion amplifier (T1.sub.1-T3.sub.1).

9. The device for detecting temporal variation of the light intensity according to claim 8, characterised in that the MOS output transistor (T5.sub.1) of the adjustable gain current mirror has: its source terminal connected to the source terminal of the MOS input transistor (T4a.sub.1); its gate terminal connected to a voltage V.sub.GA which is set by the automatic gain control circuit AGC; and, its drain terminal connected to the input of the first transimpedance amplifier (T6a.sub.1-T6d.sub.1).

10. The device for detecting temporal variation of the light intensity according to claim 7, characterised in that the MOS output transistor (T5.sub.1) of thea adjustable gain current mirror has: its source terminal connected to the source terminal of the MOS input transistor (T4a.sub.1); its gate terminal connected to a voltage V.sub.GA which is set by the automatic gain control circuit AGC; and, its drain terminal connected to the input of the first transimpedance amplifier (T6a.sub.1-T6d.sub.1).

11. The device for detecting temporal variation of the light intensity, according to claim 1, characterised in that the arbitrating and event-encoding block comprises a processor for, when the first threshold detector determines that the voltage has exceeded the higher threshold or when the second threshold detector determines that the voltage has dropped below the lower threshold, determining x and y coordinates corresponding to a pixel position in the matrix and generating an event with sign s, the sign s being determined by the first and second threshold detector, generating a word which binary-encodes the set formed by the coordinates (x,y) and the sign s.

12. The device for detecting temporal variation of the light intensity, according to claim 1, characterised in that the block for the automatic adjustment of the amplification of the photocurrent is an automatic gain control circuit AGC comprising at least: a) a replication of the collector transistor of the pixels (T4.sub.C2); b) a replication of the adjustable gain current mirror of the pixel in which the gate terminal of the MOS input transistor (T4a.sub.2) is connected to a voltage V.sub.G, its MOS output transistor (T5.sub.2) to a voltage V.sub.GA; and its output is connected to a first current reference I.sub.b1; c) a first differential voltage amplifier (A1) whose negative input is connected to the output of the mirror, whose positive input is connected to a voltage reference and whose output is connected to the output gate of the MOS transistor (T5.sub.2), generating the voltage V.sub.GA; and, d) a second differential voltage amplifier (A2), connected in unity gain configuration, which copies the voltage V.sub.GA to the gate terminals of the output transistors (T5.sub.1) of the adjustable gain current mirrors (8) of the pixels whose voltage is V.sub.GA.

13. The device for detecting temporal variation of the light intensity according to claim 12, characterised in that the block for the automatic adjustment of the amplification of the photocurrent comprises a second MOS output transistor (T5b.sub.2) of the adjustable gain mirror that shares the gate and source terminals of the first MOS output transistor (T5.sub.2), and whose drain terminal constitutes a second output from the mirror and an additional adjustment stage for each additional amplification block (10) of the pixel, where each additional adjustment stage comprises: a transimpedance amplifier (T6a.sub.2-T6d.sub.2) which is a replication of the first transimpedance amplifier in the pixels (T6a.sub.1-T6d.sub.1) whose input is connected to the output of the MOS output transistor (T5b.sub.2), thus generating a logarithmic voltage in the amplifier (T6a.sub.2 T6d.sub.2); a transconductance amplifier (T7.sub.2) which is a replication of the transconductance amplifier (11) in the additional amplification block (10) in the pixels (T7.sub.1), whose gate is connected to the output of the MOS output transistor (T5b.sub.2), its source is at a voltage V.sub.Q1 which is common to all pixels, and whose drain is connected to a current reference I.sub.b2; and, a third differential voltage amplifier (A3) whose negative input is connected to the second current reference I.sub.b2, whose positive input is connected to a voltage reference and whose output is connected to the node V.sub.Q1.

14. The device for detecting temporal variation of the light intensity, according to claim 13, characterised in that the block for the automatic adjustment of the amplification of the photocurrent comprises an additional stage of adjustment for each block of additional adjustment of the pixel, each stage of additional adjustment being connected in cascade or iteration to the previous stage of additional adjustment.

15. The device for detecting temporal variation of the light intensity, according to claim 12, characterised in that the block for the automatic adjustment of the amplification of the photocurrent comprises an additional stage of adjustment for each block of additional adjustment of the pixel, each stage of additional adjustment being connected in cascade or iteration to the previous stage of additional adjustment.

16. A method for detecting the temporal variation of the light intensity in a matrix of photosensors, which uses the device described in claim 1, characterised in that in each pixel of the matrix, the following stages are carried out: 1) transforming the light striking the pixel in a current I.sub.ph by means of the photodiode; 2) amplifying the current I.sub.ph to a value A.sub.II.sub.ph by means of the adjustable gain current mirror; 3) adapting the value A.sub.I so that the average A.sub.II.sub.ph remains constant against the temporal variations of the average brightness of all the pixels by means of the automatic amplification block of the photocurrent, which adjusts the amplification in stage 2; 4) converting the current A.sub.II.sub.ph adapted to a voltage by means of the transimpedance amplifier (T6a.sub.1-T6d.sub.1), which comprises a plurality of MOS transistors polarised in weak inversion and connected in series, each of them being connected in diode configuration; 5) determining in the switched capacitor circuit a voltage difference V=V(t.sub.2)V(t.sub.1) between two consecutive times t.sub.1 and t.sub.2, caused by a temporal variation of the light intensity and comparing the voltage difference with a fixed positive reference value V.sub.R+ and a fixed negative reference value V.sub.R, being V.sub.R+ and V.sub.R the same for all pixels of the matrix; 6) generating the digital signal s which is sent to the arbitrating and event-encoding block, the signal being selected among: a positive event every time the first threshold detector determines that the voltage exceeds the higher threshold, generated in the first threshold detector; and, a negative event every time the second threshold detector determines that the voltage drops below the lower threshold, generated in the second threshold detector; and; in the arbitrating and event-encoding block connected to the output of the pixel matrix the following stages are carried out: identifying the spatial coordinates (x,y) of the pixels of the matrix which have generated a digital signal, sending to an external device an event containing the spatial coordinates (x,y) and the signal s; and, generating a flow of events (x,y,s) representing the temporal variation of the light intensity on the matrix of photosensors.

17. The method for detecting the temporal variation of the light intensity according to claim 16, characterised in that a difference is calculated in the voltage between two reset consecutive times by the switched capacitor circuit.

18. The method for detecting the temporal variation of the light intensity according to claim 17, characterised in that, after converting the current A.sub.II.sub.ph adapted to a voltage and as a stage previous to determining in the switched capacitor circuit a voltage difference V=V(t.sub.2)V(t.sub.1), it comprises amplifying the voltage from the current conversion A.sub.II.sub.ph by means of at least one additional amplification block (10).

19. The method for detecting the temporal variation of the light intensity according to claim 16, characterised in that, after converting the current A.sub.II.sub.ph adapted to a voltage and as a stage previous to determining in the switched capacitor circuit a voltage difference V=V(t.sub.2)V(t.sub.1), it comprises amplifying the voltage from the current conversion A.sub.II.sub.ph by means of at least one additional amplification block (10).

Description

DESCRIPTION OF THE FIGURES

(1) FIG. 1shows an exemplary embodiment of the sensor of the photodiode matrix for detecting Lichtsteiner's time-dependent visual scenes described in U.S. Pat. No. 7,728,269 B2 and pertaining to the state of the art.

(2) FIG. 2shows an exemplary embodiment of the integrated circuit device for detecting temporal variation of the light intensity in a photosensor matrix object of the present invention.

(3) FIG. 3shows a particular embodiment of the pixels making up the matrix of pixels of the sensor shown in FIG. 2.

(4) FIG. 4shows another particular embodiment of the pixels making up the matrix of pixels of the sensor shown in FIG. 2, wherein said pixel has an additional amplification block.

(5) FIG. 5shows an exemplary embodiment of the automatic gain control circuit of the sensor shown in FIG. 2.

(6) FIG. 6shows an exemplary embodiment of the automatic gain control circuit of the sensor shown in FIG. 2, wherein said circuit has two additional adjusting stages.

(7) FIG. 7shows the flowchart of a particular embodiment of the method object of the present invention using the device shown in FIG. 2.

(8) FIG. 8shows a transimpedance amplifier with N.sub.1 MOS transistors polarized in weak inversion, in diode configuration and connected in series of those used in the present invention. Such transistors are described in patent ES 201130862.

EXAMPLES

(9) Next, an illustrative and non-limiting description is made of several examples of particular embodiments of the invention by making reference to the numbering adopted in the figures.

(10) In a first example of an embodiment of the device of the present invention, FIG. 2 shows an integrated circuit device for detecting temporal variation of the light intensity in a matrix of photosensors. Said device consists of a two dimensional matrix (1) (such matrix could also be one dimensional) of pixels which in turn consists of a plurality (x, y) of pixels (6), an arbitrating and event-encoding block (2), which also communicates events to the exterior of the apparatus connected to each and every one of the pixels (6), and a block for the automatic current gain adjustment AGC (3) connected to the matrix (1). Said connection between the matrix (1) and the AGC block (3) is performed by interposing a MOS transistor (4) and a capacitor connected to a voltage V.sub.avg setting a representation of the space-time average of the photocurrents generated in the matrix (1) of pixels, thus obtaining, at the output of the AGC block (3), the voltages V.sub.GA and V.sub.Qi, wherein i varies from 1 to n and n is the total number of transimpedance amplification blocks used by the pixels (6).

(11) FIGS. 3 and 4 show two particular embodiments of one pixel (6) forming the matrix of pixels (1) of FIG. 2. In both preferred embodiments, the pixel (6) comprises a photodiode D.sub.1, two capacitors C.sub.3 and C.sub.4, and a series of labelled transistors Ti, where the index i takes the numerical values of 1 to 15, plus optionally letters a, b, c, or d. The photodiode D.sub.1 provides a photocurrent I.sub.ph1 proportional to the light striking the pixel (6). The transistors T1.sub.1 to T3.sub.1 provide a voltage amplifier (7) with input V.sub.1 and output V.sub.2, which is connected to the sources of the PMOS transistors T4a.sub.1, T4b.sub.1 and T5.sub.1. This voltage amplifier (7) together with the transistor T4a.sub.1 constitute the input branch of a current mirror (8) which receives the photocurrent I.sub.ph1 as an input while it achieves the setting of the voltage V.sub.1 at a constant value. The current mirror (8) has two output branches formed by the transistors T4b.sub.1 and T5.sub.1. The output branch formed by the transistor T4b.sub.1 presents unity gain, as T4b.sub.1 shares the gate voltage with the input branch transistor T4a.sub.1 and transistors T4a.sub.1 and T4b.sub.1 are made with the same size. Optionally, such unity gain could be changed into a higher or lower one, either by changing the size ratios among T4a.sub.1 and T4a.sub.2, or by connecting them to different gate voltages, if considered convenient for design considerations. Without loss of generality, it has been herein considered that the gain is the unity. Therefore, T4b.sub.1 provides a copy of the photocurrent I.sub.ph1. This current is sent to the transistor T4c.sub.1, which is connected in diode configuration between the nodes V.sub.s1 and V.sub.avg, both being shared by all the pixels of the matrix. Thus, in the shared node V.sub.avg a voltage, which depends on the photocurrent average of all the pixels, is formed. The transistor T5.sub.1 forming the second output branch of the current mirror provides an amplified current A.sub.II.sub.ph1, wherein the current amplification is determined by the difference between the gate voltages V.sub.G and V.sub.GA. This voltage difference, which is common for all the pixels (6) as they all share these two voltages, is set by the automatic gain control block (3) described below. The so amplified photocurrent A.sub.II.sub.ph1 is connected to a transimpedance amplifier formed by the transistors T6a.sub.1 to T6d.sub.1, each in diode configuration, and which must be polarized in weak inversion. The voltage V.sub.o1 is the output voltage of the transimpedance stage and presents a value V.sub.o1=N.sub.1V.sub.o log (A.sub.II.sub.ph1/I.sub.s), wherein, in this particular example of a mode for carrying out the invention, the number of transistors in the transimpedance amplifier is N.sub.1=4. In the mode for carrying out the invention shown in FIG. 3, this output voltage V.sub.o1 is connected to the input of the switched capacitor stage (14) formed by the capacitors C3 and C4 and the transistors T10a.sub.1, T10b.sub.1 and T11.sub.1.

(12) The switched capacitor circuit (14) comprising the capacitors C3 and C4 and the transistors T10a.sub.1, T10b.sub.1 and T11.sub.1, copies to V.sub.diff1 the voltage variation at V.sub.o1 from a previous reset time t.sub.1, multiplied by the capacitive gain A.sub.c1=C4/C3. Thus, V.sub.diff1(t)=A.sub.c1(V.sub.o1(t)V.sub.o1(t.sub.1))=A.sub.c1N.sub.1V.sub.o log (I.sub.ph1(t)/I.sub.ph1 (t.sub.1)). Note that, in this expression, all parameters liable to undergo large dispersions from pixel to pixel have disappeared, remaining only the capacitive amplification A.sub.c1, which presents low dispersion, the numbers N.sub.1 with no dispersion and the physical parameter V.sub.o having low dispersion. The transistors T12.sub.1 to T13.sub.1 detect whether V.sub.diff1 exceeds a specific positive threshold V.sub.R+ and if so, it generates a positive event (ON). The transistors T14.sub.1 to T15.sub.1 detect whether V.sub.diff1 drops below a negative threshold V.sub.R and if so, they generate a negative event (OFF). Every time the pixel (6) generates an event, a reset of capacitor C3 occurs by means of the reset transistor T11.sub.1. Thus, the pixel (6) generates a positive event at the time t.sub.2 if V.sub.R+=A.sub.c1N.sub.1V.sub.o log (I.sub.ph1(t.sub.2)/I.sub.ph1(t.sub.1)), and a negative event if V.sub.R=A.sub.c1N.sub.1V.sub.o log (I.sub.ph1(t.sub.2)/I.sub.ph1(t.sub.1)). This can also be expressed as I/I=exp ((V.sub.R+/)/(A.sub.c1N.sub.1V.sub.o))1=/.

(13) In the exemplary embodiment of the pixel shown in FIG. 4, it was chosen to add a second amplifier stage by adding one additional amplification block (10). This requires adding a transconductance stage (11), a current mirror (12) and a second transimpedance amplifier (13). The transconductance stage (11) constitutes the MOS transistor T7.sub.1 polarized in weak inversion, which provides a current I.sub.2=I.sub.s exp ((V.sub.o1V.sub.Q1)/V.sub.o). The current mirror (12) is made up, in this case and without loss of generality, of the transistors T8a.sub.1 to T8c.sub.1 and they copy the current I.sub.2, which is present in the input branch, to the output branch. The gain or attenuation in this copy process is given by the relative proportion in the sizes of the transistors T8b.sub.1 and T8c.sub.1. Without loss of generality, it has been considered that transistors T8b.sub.1 and T8c.sub.1 are the same size, so that the gain of the mirror (12) will be the unity. Thus, the mirror (12) provides a current equal to I.sub.2. This current enters in a second transimpedance stage (13) made up, in this case, of three transistors, namely T9a.sub.1, T9b.sub.1 and T9c.sub.1, which provide an output voltage V.sub.o2=N.sub.1N.sub.2V.sub.o log (A.sub.ph1/I.sub.s), wherein, in this particular exemplary embodiment, the number of transistors in the second transimpedance amplifier (13) is N.sub.2=3. The additional amplification block (10) made up of the transistors T7.sub.1, T8i.sub.1, T9j.sub.1, can be repeated as many times as needed and viable in order to increase the amplification factor in the final output voltage of the last transimpedance stage. This output is connected to the switched capacitor circuit (14). FIG. 4 shows an example in which the number of transimpedance amplifiers is n=2 because there is only one additional amplification block (10), and therefore the output of the last stage is V.sub.o2. However, by placing more additional amplification blocks in cascade or iteration in the output of this first additional amplification block, where the last of these blocks is connected to the input of the switched capacitor circuit (14), an increase of the amplification factor is achieved in the final output voltage of the last transimpedance stage (13). Thus, in the output of the transimpedance stage (13) of the last additional amplification block (10), a voltage V.sub.on is obtained (n=number of additional amplification blocks positioned in cascade or iteration minus one, or n=number of transimpedance amplifiers). So in this case V=.sub.diff1=A.sub.C1 (V.sub.o2(t)V.sub.o2(t.sub.1))=A.sub.c1N.sub.1N.sub.2V.sub.o log (I.sub.ph1/I.sub.ph1 (t.sub.1)), and as in the previous way, the pixel generates a positive event in the time t.sub.2 if V.sub.R+=A.sub.c1N.sub.1N.sub.2V.sub.o log (I.sub.ph1 (t.sub.2)/I.sub.ph1 (t.sub.1)), and a negative event if V.sub.R=A.sub.c1N.sub.1N.sub.2V.sub.o log (I.sub.ph1 (t.sub.2)/I.sub.ph1 (t.sub.1)). In this way a sensitivity to the contrast .sub.+/=exp ((V.sub.R+/)/(A.sub.c1N.sub.TV.sub.o))1 is obtained, where N.sub.T=N.sub.1N.sub.2. In one embodiment with n transimpedance amplifiers, it would be N.sub.T=N.sub.1N.sub.2 . . . N.sub.n.

(14) Therefore, if in FIG. 1 a ratio between C2 and C1 of value A.sub.c=C2/C1=20 was adjusted, in the exemplary circuits in FIGS. 3 and 4 A.sub.c1N.sub.1N.sub.2=24 can be achieved by making A.sub.c1=2 (with N.sub.1=4, N.sub.2=3), which is achieved with the capacitors occupying very little area in an integrated circuit embodiment. In a typical embodiment, A.sub.c1=5 would be set, which also translates into an insignificant area consumption within the pixel (6) while achieving a quite higher total gain A.sub.c1N.sub.1N.sub.2=60, with consequent significant improvement of the contrast sensitivity, which, under these circumstances can be set at about 1%.

(15) On the outside of the two-dimensional matrix (1) of pixels shown in FIG. 2 there is an automatic gain control circuit (3), of which two exemplary embodiments are shown in FIGS. 5 and 6. This circuit shares, with all the pixels (6), the nodes referred to as V.sub.avg, V.sub.G, V.sub.GA and V.sub.Qi, where i=1 to k, where k-1 is the number of additional amplification blocks (10) that have been included in the pixels (6), being in turn k=n1. The node V.sub.avg is a representation of the spatial-temporal averaging <I.sub.ph> of the photocurrent received by all photodiodes D.sub.1 of the matrix of pixels (1). This voltage controls the gate of the transistor T4c.sub.2, thus generating a current equal to the spatio-temporal average <I.sub.ph>. Therefore, the transistor T4c.sub.2 is acting as a photodiode that provides the average photocurrent <I.sub.ph>. The transistors T1.sub.2, T2.sub.2 and T3.sub.2 do the same function as T1.sub.1, T2.sub.1 and T3.sub.1 in FIGS. 3 and 4 within each pixel, that is, they form a voltage amplifier (15). The transistors T4a.sub.2 and T5.sub.2 do the same function as T4a.sub.1 and T5.sub.1 within each pixel (6), that is, they form an adjustable gain current mirror (16), said gain depending on the difference of voltages V.sub.GAV.sub.G. The output of the current mirror (16) which corresponds to the MOS transistor T5.sub.2 is sent to a source of current reference with a value I.sub.b1. The differential voltage amplifier A1 is connected so that it compares the voltage in the output of the adjustable gain current mirror (16) with a voltage reference, and its output adjusts the gate of the output transistor T5.sub.2 of the adjustable gain current mirror, i.e., it controls the voltage V.sub.GA. The result achieved with this amplifier A1 so connected is that the gain of the current mirror (16) formed by the transistors T4a.sub.2 and T5.sub.2, is self-adjusted, so that A.sub.1<I.sub.ph> equals I.sub.b1. The gate voltage V.sub.GA so generated is copied to the gates of the transistors T5.sub.1 of all pixels as the voltage V.sub.GA by means of the differential voltage amplifier set in unity gain A2. If the pixels contain a single transimpedance stage, i.e., if n=1, the automatic gain control circuit would end here (FIG. 5).

(16) If the pixels contain a second transimpedance stage, i.e., an additional first amplification block (10), i.e., if n=2, then, the transistor T5b.sub.2, which provides an additional copy of the output of the adjustable gain current mirror (16), providing a current A.sub.I<I.sub.ph>, and an additional first adjustment stage (17) should be added. This circuit would then comprise a transimpedance amplifier (18), a transconductance amplifier (19), a current reference I.sub.b2 and a differential voltage amplifier A3. The current A.sub.I<I.sub.ph> is provided to the transimpedance amplifier (18) formed by the transistors T6a.sub.2 to T6d.sub.2, which are a replication of the transistors T6a.sub.1 to T6d.sub.1 in FIG. 3, which form the first transimpedance stage (9) in the pixels (6). The output of this transimpedance stage is connected to the transconductance amplifier (19) formed by the transistor T7.sub.2, which is a replication of the transistor T7.sub.1 in all pixels (6). The output of the transconductance amplifier (19) is connected to a current reference I.sub.b2. This output is also connected to the input of a differential voltage amplifier, which compares it with a reference voltage and provides its output to the node V.sub.Q1 of the transductance amplifier. The result of this configuration is that the voltage V.sub.Q1 is self-adjusted so that the transductance amplifier (19) T7.sub.2 provides the current I.sub.b2. Since the voltage V.sub.Q1 is shared with all pixels (6) of the matrix (1), it is achieved that the transconductance amplifiers (11) T7.sub.1 of all pixels (6) operate at an average current equals to I.sub.b2.

(17) If the pixels comprise a third transimpedance stage, that is, an additional second amplification block (10) connected in cascade or iteration to the first one, i.e., if n=3, an additional second adjustment stage (20) should be added to the automatic gain control circuit (3). This exemplary embodiment is shown in FIG. 6. This would contain a replication of the second transimpedance stage (21) made up of the transistors T9a.sub.2, T9b.sub.2 and T9c.sub.2 supplied by a current equal to the average of the corresponding current in the pixels (6). In this particular example, this current would be equal to I.sub.b2, since the mirror formed by T8a.sub.1 to T8c.sub.1 in the pixels (6) is supposed to be of unity gain. If their gain were not unity, this current I.sub.b2 should be multiplied by said gain. The transconductance amplifier (22) T10.sub.2 and the amplifier A4 together with a current reference I.sub.b3, which represents the value of the average current to which it is desired to make operate the third transconductance amplifier within the pixels, are also added. As in the additional first adjustment stage (17), in the additional second adjustment stage (20) a voltage V.sub.Q2 is generated, which is shared with all pixels (6), so adjusting the average current of the second transconductance amplifier (13) in the pixels (6).

(18) If the pixels (6) have more additional amplification blocks (10), more additional adjustment stages (20) arranged in cascade or iteration would be repeated in the automatic gain control circuit (3).

(19) FIG. 7 shows an exemplary embodiment of the method object of the present invention. Said method is shown by a flow chart comprising two parts, a first part (45), which describes the sequence of stages to be performed within each pixel, and the second part (44), which describes the stages to be performed outside the matrix of pixels to perform the automatic adjustment of the current gain. Thus, firstly in each of the pixels, the integrated light sensor provides (23) a photocurrent I.sub.ph, which is proportional to the light striking the pixel at each time. Next, a copy of the photocurrent is sent (24) to the block for the automatic gain adjustment (AGC). This photocurrent is amplified (25) thus becoming A.sub.II.sub.ph, where the current gain A.sub.I is determined by the AGC itself. The resulting current A.sub.II.sub.ph is converted into a voltage (26) by a transimpedance amplifier (pertaining to the state of the art) of N.sub.1 MOS transistors (43) polarised in weak inversion in diode configuration and connected in series, as shown in FIG. 8. Each MOS transistor in diode configuration generates a potential difference value V.sub.o log(A.sub.I/I.sub.ph/I.sub.s), wherein V.sub.o is a physical parameter that undergoes low dispersion from pixel to pixel, and I.sub.s is a technological parameter that undergoes a significant dispersion from pixel to pixel. Consequently, the output voltage of the transimpedance stage will be V.sub.o1=N.sub.1V.sub.o log (A.sub.II.sub.ph/I.sub.s).

(20) Depending on each case, the voltage V.sub.o1 obtained is evaluated (27) to decide whether it is sufficient or not, so that in the case that more amplification were not needed, the output voltage V.sub.o1 will be used directly in the stage (33) shown below. If more amplification were required, the output voltage V.sub.o1 is transformed (28) into a current I.sub.2=I.sub.o2 exp(V.sub.o1/V.sub.o) by means of a transconductance amplifier. Said current I.sub.2 is copied (29) with an optional amplification or attenuation A.sub.2, resulting in a current A.sub.2I.sub.2. If A.sub.2=1, there is no amplification or attenuation. If A.sub.2>1, there is amplification, and if A.sub.2<1, there is attenuation. This gain/attenuation A.sub.2 does not require automatic gain adjustment. The resulting current A.sub.2I.sub.2 is converted into a voltage (30) by a transimpedance amplifier of N.sub.2 MOS transistors (43) polarised in weak inversion, in diode configuration, and connected in series, similarly as shown in FIG. 8. The output voltage of this transimpedance stage will be V.sub.o2=N.sub.2V.sub.0 log(A.sub.2I.sub.2/I.sub.s).

(21) The stages (27-30) can be repeated again n2 times, if the voltage V.sub.o2 in the output were still not sufficient (31). In the end, the resulting output voltage will be V.sub.on=N.sub.1N.sub.2 . . . N.sub.nV.sub.0 log (A.sub.1A.sub.2 . . . A.sub.nI.sub.ph/I.sub.s). Subsequently, the difference between the resulting voltage output V.sub.on (t) and that in the immediately preceding reset time t.sub.reset is calculated (33). In this way, the value V (t)=V.sub.on(t)V.sub.on(t.sub.reset)=N.sub.1N.sub.2 . . . N.sub.nV.sub.o log(I.sub.ph(t)/I.sub.ph(t.sub.reset)), is obtained, wherein the parameters with high dispersion A.sub.i and I.sub.s have disappeared. If in a given time V (t) were higher than a pre-set positive voltage reference V.sub.R+ (34) the next reset time is established, so updating t.sub.reset=t (35), and the pixel outputs a positive event (36). If in a given time V(t) drops below a pre-set negative voltage reference V.sub.R (37), the next reset time is established, so updating t.sub.reset=t (38), and the pixel outputs a negative event (39). Finally, for each event generated by each pixel, an event (40) is sent to the outside of the sensor formed by the coordinates (x, y) of the pixel that has generated the event as well as the sign s of the generated event.

(22) As for the second part of the method, the average photocurrent <I.sub.ph> (41) is calculated in the AGC by using the photocurrent copies provided by all pixels. Then, the quotient A.sub.II.sub.b1/<I.sub.ph> is calculated (42) where I.sub.b1 is the average current level at which it is desired to make operate the first transimpedance amplifier, and this resulting value is the one used as the current amplification gain in all pixels, as described in the photocurrent amplification stage (25) in the AGC, which becomes A.sub.II.sub.ph.