A method and apparatus for dynamic edge imaging with embedded probability coding

Abstract

The application provides a dynamic edge imaging method and device embedded with probability coding. In the case of collecting an image to be processed, an electrical signal corresponding to the image to be processed is acquired in real time, the electrical signal is selected and amplified, and a current voltage is obtained based on the selected and amplified electrical signal. The current voltage is the voltage corresponding to a single pixel unit. The current voltage includes multiple and corresponding pixel units that are different. Based on the relationship between a predetermined current reference voltage, a voltage to be compared, and the current voltage, a pulse signal is obtained. The current reference voltage and the current voltage have a corresponding relationship. Based on the pulse signal and a pixel coordinate obtained in advance, a to-be-recognized image is generated. In this way, pulsed image data can be obtained, the defect of delay in pulsed processing of image data is solved, and the user experience can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to a dynamic edge imaging method and device embedded with a probability coding. BACKGROUND

[0002] Under the background of the big data era, the classic architecture of separating the memory and the processor in the traditional data processing method brings the Von Neumann bottleneck problem, and this processing method has gradually failed to meet people's growing data processing needs. Compared with this, the information activity of the human brain nervous system has the characteristics of large-scale parallelism, distributed storage and processing, self-organization, self-adaptation and self-learning, and the information storage and processing has no obvious boundary. Therefore, in recent years, brain-like computing has gradually attracted the attention of many scholars, trying to model the brain from the micro to the macro level, and further promoting the current artificial intelligence process based on deep learning technology.

[0003] The spiking neural network (SNN, Spiking Neuron Networks) as the third generation of neural networks designed to simulate the mechanism of biological neurons has more brain-like characteristics in connection mode, information processing mechanism and synaptic weight learning method. The simulated neurons in the spiking neural network are closer to the actual characteristics: neurons will only be activated when the membrane potential reaches the threshold, and the spiking neural network considers the pulse generation time, enhancing the ability to process spatio-temporal data.

[0004] The spiking neural network processes pulse time series, and there are two ways to obtain the pulse sequence: the first way is to convert the image into a pulse sequence according to a certain coding method (for example: threshold coding method, Gaussian difference method, frequency coding method, etc.), but the above methods have the problems of serious information loss and low coding efficiency. The second way is derived from the neuromorphic vision sensor. However, at present, due to the immaturity of the principle and technology of the neuromorphic vision sensor, the high price and other reasons, the application of the neuromorphic vision sensor is not universal.

[0005] In addition, even the spiking neural network as the third generation of neural networks still needs a lot of time for model construction and training. For example, in specific application scenarios such as autonomous driving, images need to be generated in time and sent to the computing system for processing. If the image is obtained first and then sent to the backend computing platform for pulse processing, the computing power of the computing platform is consumed, and the recognition speed of the spiking neural network is reduced, that is, there is a delay in the process of pulse processing of image data, which reduces the user's experience. SUMMARY

[0006] The application provides a dynamic edge imaging method and device embedded with probability coding, which aims to solve the delay defect in the process of pulse processing of image data in the prior art, and improve the user experience.

[0007] The application provides a dynamic edge imaging method embedded with probability coding, which comprises the following steps:

[0008] In the case of collecting an image to be processed, an electrical signal corresponding to the image to be processed is acquired in real time;

[0009] The electrical signal is selected and amplified, and a current voltage is obtained based on the selected and amplified electrical signal, wherein the current voltage is the voltage corresponding to a single pixel unit, and the current voltage includes multiple and corresponding pixel units which are different;

[0010] Based on the relationship between a predetermined current reference voltage, a predetermined comparison voltage and the current voltage, a pulse signal is obtained, wherein the current reference voltage and the current voltage have a corresponding relationship;

[0011] Based on the pulse signal and a pixel coordinate obtained in advance, a to-be-recognized image is generated.

[0012] According to the dynamic edge imaging method embedded with probability coding provided by the application, the comparison voltage includes a driving voltage of a transistor and a voltage division voltage, and the voltage division voltage is a voltage obtained based on a power supply voltage and a magnetic tunnel junction voltage.

[0013] The step of obtaining the pulse signal based on the relationship between the predetermined current reference voltage, the predetermined comparison voltage and the current voltage comprises the following steps:

[0014] Based on the characteristics of the transistor, the current voltage and the driving voltage, it is determined whether the transistor is turned on or not;

[0015] In the case that the transistor is turned on, the voltage division voltage and the current reference voltage are compared;

[0016] In the case that the voltage division voltage is greater than the current reference voltage, a high-level signal is outputted;

[0017] In the case that the voltage division voltage is not greater than the current reference voltage, a low-level signal is outputted;

[0018] In the case that the transistor is not turned on, the power supply voltage and the current reference voltage are compared;

[0019] In the case that the power supply voltage is greater than the current reference voltage, a high-level signal is outputted;

[0020] Output a low-level signal in a case where the power supply voltage is not greater than the current reference voltage.

[0021] The method for dynamic edge imaging with embedded probability coding provided by the application further comprises the following steps before the step of obtaining the pulse signal based on the relationship between the predetermined current reference voltage, the pre-acquired voltage to be compared and the current voltage.

[0022] In a case where the electrical signal meets preset sampling conditions, the electrical signal is randomly sampled to obtain a current reference voltage corresponding to the electrical signal.

[0023] The step of randomly sampling the electrical signal to obtain a current reference voltage corresponding to the electrical signal comprises the following steps.

[0024] A preset number of electrical signals are randomly sampled from the electrical signal as sampled electrical signals.

[0025] The current reference voltage is determined based on the intensity values of the sampled electrical signals.

[0026] The electrical signal is a plurality of pixel units corresponding electrical signals.

[0027] The step of gating and amplifying the electrical signal and obtaining a current voltage based on the current gated and amplified electrical signal comprises the following steps.

[0028] The corresponding electrical signal is selected from the electrical signal for gating and amplification according to a preset selection mode to obtain a voltage to be processed, wherein the preset selection mode comprises a row-by-row selection mode and a column-by-column selection mode.

[0029] The voltage corresponding to each pixel unit included in the voltage to be processed is sequentially taken as a current voltage.

[0030] The step of generating an image to be recognized based on the pulse signal and pre-acquired pixel coordinates comprises the following steps.

[0031] The pulse signal is converted into a digital signal, the pixel coordinates corresponding to the digital signal are added to the digital signal, and an image to be recognized is generated based on the digital signal to which the pixel coordinates are added.

[0032] The step of generating an image to be recognized based on the digital signal to which the pixel coordinates are added comprises the following steps.

[0033] The digital signal after adding the pixel coordinates is sorted to generate a to-be-recognized image.

[0034] According to the application, a dynamic edge imaging method with embedded probability coding is provided.

[0035] The to-be-recognized image is input into a pre-trained pulse neural network for image recognition to obtain a recognition result.

[0036] The application further provides a dynamic edge imaging device with embedded probability coding.

[0037] An image acquisition unit is configured to acquire an electrical signal corresponding to a to-be-processed image in real time when the to-be-processed image is collected.

[0038] A gating and amplifying unit is configured to gate and amplify the electrical signal and obtain a current voltage based on the electrical signal after gating and amplifying, wherein the current voltage is a voltage corresponding to a single pixel unit, and the current voltage includes a plurality of different corresponding pixel units.

[0039] A pulse signal acquisition unit is configured to acquire a pulse signal based on a relationship among a pre-determined current reference voltage, a pre-acquired to-be-compared voltage, and the current voltage, wherein the current reference voltage has a corresponding relationship with the current voltage.

[0040] A generating unit is configured to generate a to-be-recognized image based on the pulse signal and a pre-acquired pixel coordinate.

[0041] According to the application, a dynamic edge imaging device with embedded probability coding is provided.

[0042] The current voltage is an input voltage of a gate of the transistor, a drain level of the transistor is connected with one end of the magnetic tunnel junction, the other end of the magnetic tunnel junction is connected with a negative electrode of the power supply, a positive electrode of the power supply is connected with a source electrode of the transistor, and the current reference voltage is an input voltage of a positive input end of the comparator.

[0043] The to-be-compared voltage includes a driving voltage of the transistor and a voltage division voltage, the voltage division voltage is a voltage acquired based on a supply voltage and a magnetic tunnel junction voltage, and the supply voltage is a voltage provided by the power supply.

[0044] The pulse signal is output by the comparator based on the relationship among the current reference voltage, the to-be-compared voltage, and the current voltage.

[0045] The application provides a dynamic edge imaging method and device embedded with probability coding, which comprises the following steps: acquiring an electric signal corresponding to a to-be-processed image in a case that the to-be-processed image is collected, carrying out gating and amplification processing on the electric signal, and obtaining a current voltage based on the current electric signal after gating and amplification, wherein the current voltage is a voltage corresponding to a single pixel unit, the current voltage comprises a plurality of different corresponding pixel units, and the relationship among a current reference voltage, a to-be-compared voltage and the current voltage is determined in advance, the pulse signal is acquired based on the relationship, and the to-be-identified image is generated based on the pulse signal and a pixel coordinate acquired in advance.

[0046] In this way, the pulse signal can be acquired based on the relationship among the current reference voltage, the to-be-compared voltage and the current voltage in a case that the to-be-processed image is collected, and the to-be-identified image can be acquired based on the pulse signal, that is, the pulsed image data can be acquired, the pulsed image data is acquired in the process of collecting the to-be-processed image, the defect that there is a delay in the pulsed processing of the image data is solved, and the use experience of a user can be improved. BRIEF DESCRIPTION OF DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can be obtained by those skilled in the art without any creative effort on the basis of these drawings.

[0048] Figure 1 is one of the flowcharts of the dynamic edge imaging method embedded with probability coding provided by the application;

[0049] Figure 2 is a schematic diagram of the to-be-processed image provided by the application;

[0050] Figure 3 is a schematic diagram of the to-be-identified image provided by the application;

[0051] Figure 4 is one of the structural schematic diagrams of the dynamic edge imaging device embedded with probability coding provided by the application;

[0052] Figure 5 is the structural schematic diagram of the random circuit provided by the application;

[0053] Figure 6 is the schematic diagram of the pulse signal provided by the application;

[0054] Figure 7 is a structural schematic diagram of a dynamic edge imaging device with embedded probability coding provided by the present application.

[0055] Reference signs:

[0056] 510: transistor; 520: magnetic tunnel junction; 530: comparator; 701: color filter; 702: pixel sensor; 703: gate; 704: sampling and averaging device; 705: amplifier; 706: position encoder; 707: timing circuit; 708: random circuit; 709: buffer; 710: calculation module. DETAILED DESCRIPTION

[0057] In order to make the objects, technical solutions and advantages of the present application clearer, the technical solutions in the present application will be described clearly and completely below in combination with the drawings in the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0058] In order to reduce the delay caused by the pulsing processing of image data and improve the user experience, the present application provides a dynamic edge imaging method and device with embedded probability coding, which will be described below in combination with the drawings. Figure 1 The present application provides a dynamic edge imaging method and device with embedded probability coding, which will be described below in combination with the drawings.

[0059] As shown in the drawings, the present application provides a dynamic edge imaging method and device with embedded probability coding, which will be described below in combination with the drawings. Figure 1 The present application provides a dynamic edge imaging method and device with embedded probability coding, which will be described below in combination with the drawings.

[0060] S101, in the case of acquiring a to-be-processed image, the electrical signal corresponding to the to-be-processed image is acquired in real time.

[0061] In order to acquire the to-be-processed image, an image acquisition device can be used to acquire the to-be-processed image, wherein the image acquisition device can be a CMOS (Complementary Metal Oxide Semiconductor) image sensor, and of course other image sensors (pixel sensors) can also be set according to actual use requirements, which are not limited here.

[0062] Since the image acquisition device has a corresponding exposure duration when acquiring the to-be-processed image, the electrical signal corresponding to the to-be-processed image is not the same for each time point in the exposure process. In other words, the electrical signal (also referred to as a pixel signal) corresponding to the to-be-processed image is not the same for each time point at which the to-be-processed image is acquired, and the electrical signal corresponding to each pixel unit included in the to-be-processed image is also not the same.

[0063] Therefore, in order to be able to implement the pulsing processing of the to-be-processed image during the acquisition of the to-be-processed image, the electrical signal corresponding to the to-be-processed image can be acquired in real time when the to-be-processed image is acquired, wherein the electrical signal is the electrical signal corresponding to the pixel array.

[0064] Taking a CMOS image sensor as an example, the above step S101 is described: Since the CMOS image sensor does not have color resolution capability, in order to be able to acquire the to-be-processed image, visible light can be filtered to enable the pixel units corresponding to the CMOS image sensor to distinguish three basic colors of red, green, and blue, so as to achieve real-time acquisition of the electrical signal corresponding to the to-be-processed image, that is, to achieve acquisition of the electrical signal corresponding to the pixel array, the pixel array including a plurality of pixel units, when the to-be-processed image is acquired.

[0065] In an embodiment, the CMOS image sensor can be compatible with a color filter arrangement device of a common image format, so that the color filter arrangement device (color filter) can be used to filter visible light, for example, the color filter arrangement device can be arranged in front of the CMOS image sensor. Wherein the common image format can include: Bayer, RGBE, RGBW, CYYM, CYGM, etc., which are all reasonable and are not limited here.

[0066] S102, the electrical signal is gated and amplified, and a current voltage is obtained based on the current gated and amplified electrical signal.

[0067] After acquiring the electrical signal corresponding to a certain time point, since the electrical signal corresponding to the pixel unit is small, it cannot meet the preset voltage requirement, wherein the preset voltage requirement is the voltage requirement for subsequent acquisition of the pulse signal.

[0068] Therefore, in order to facilitate subsequent acquisition of the pulse signal, the electrical signal can be gated and amplified, and a current voltage can be obtained based on the current gated and amplified electrical signal, wherein the current voltage is the voltage corresponding to a single pixel unit, and the current voltage includes a plurality of different corresponding pixel units.

[0069] In an embodiment, the electrical signal is an electrical signal corresponding to a plurality of pixel units, and after the electrical signal is acquired, electrical signals corresponding to a preset number of pixel units can be selected for amplification, so as to subsequently obtain the current voltage. The number of selected electrical signals of the pixel units is less than the number of acquired electrical signals of the pixel units.

[0070] As an embodiment, the step of selecting and amplifying the electrical signal can be performed multiple times to realize traversal of electrical signals corresponding to each pixel unit included in the to-be-processed image. That is, the current selected and amplified electrical signal includes multiple electrical signals, and each current selected and amplified electrical signal is different.

[0071] S103, acquiring a pulse signal based on a relationship between a predetermined current reference voltage, a to-be-compared voltage acquired in advance, and the current voltage.

[0072] After the current voltage is acquired, a pulse signal can be acquired based on a relationship between a predetermined current reference voltage, a to-be-compared voltage acquired in advance, and the current voltage. The current reference voltage has a corresponding relationship with the current voltage, the current reference voltage is a reference voltage determined based on the to-be-processed image, and the to-be-compared voltage includes a voltage determined based on an electrical signal corresponding to the magnetic tunnel junction. The pulse signal includes multiple pulse signals, and the pulse signal has a corresponding pixel unit.

[0073] S104, generating a to-be-recognized image based on the pulse signal and a pixel coordinate acquired in advance.

[0074] After the pulse signal is acquired, a to-be-recognized image can be generated based on the pulse signal and a pixel unit acquired in advance, that is, pulse image data is generated, which realizes acquisition of pulse image data in the process of collecting the to-be-processed image, solves the defect of delay in pulse processing of image data, and can improve the user experience.

[0075] As an embodiment of the present application, the to-be-compared voltage can include a driving voltage of a transistor and a divided voltage, and the divided voltage is a voltage acquired based on a supply voltage and a magnetic tunnel junction voltage.

[0076] The transistor can be an NMOS (Negative channel Metal Oxide Semiconductor, N-type metal oxide semiconductor) transistor or a PMOS (Positive channel Metal Oxide Semiconductor, P-type metal oxide semiconductor) transistor.

[0077] A magnetic tunnel junction (MTJ) is a sandwich structure device, including a reference layer, a tunnel barrier layer and a free layer from top to bottom. The free layer and the fixed layer are two layers of magnetic thin films, forming a nanometer magnet, and the insulating layer is an oxide film, which separates the electron migration in the free layer and the fixed layer as a tunnel barrier. By applying an external electromagnetic field control, the magnetic moment direction of the free layer can be switched to be parallel or anti-parallel to the fixed layer. If it is parallel, the resistance of the magnetic tunnel junction is low, and the resistance is high in the anti-parallel state, so that the resistance difference can be used as a logic state. When the tunnel barrier in the magnetic tunnel junction is low, the parallel state and the anti-parallel state appear random jump phenomenon due to thermal noise, generating random output characteristics.

[0078] The step of obtaining the pulse signal based on the relationship between the predetermined current reference voltage, the pre-acquired voltage to be compared and the current voltage can include:

[0079] Determining whether the transistor is turned on based on the characteristics of the transistor, the current voltage and the driving voltage.

[0080] In the case of an NMOS transistor, if the current voltage is greater than the driving voltage, it is determined that the transistor is turned on, and if the current voltage is not greater than the driving voltage, it is determined that the transistor is not turned on.

[0081] In the case of a PMOS transistor, if the current voltage is not greater than the driving voltage, it is determined that the transistor is turned on, and if the current voltage is greater than the driving voltage, it is determined that the transistor is not turned on.

[0082] In the case that the transistor is turned on, the divided voltage and the current reference voltage are compared.

[0083] In the case that the transistor is determined to be turned on, the divided voltage and the current reference voltage can be compared to determine the level signal output subsequently. In the case that the divided voltage is greater than the current reference voltage, a high level signal is output. In the case that the divided voltage is not greater than the current reference voltage, a low level signal is output.

[0084] In the case that the transistor is not turned on, the supply voltage and the current reference voltage are compared.

[0085] In the case that the transistor is determined to be not turned on, the supply voltage and the current reference voltage can be compared to determine the level signal output subsequently. In the case that the supply voltage is greater than the current reference voltage, a high level signal is output. In the case that the supply voltage is not greater than the current reference voltage, a low level signal is output.

[0086] As can be seen, in this embodiment, it can be first determined whether the transistor is conducting, and different methods can be used to determine the output level signal according to the conduction status of the transistor, thereby obtaining the pulse signal.

[0087] As one embodiment of the present invention, before the step of obtaining the pulse signal based on the relationship between the predetermined current reference voltage, the pre-acquired comparison voltage, and the current voltage, the method may further include:

[0088] When the electrical signal meets preset sampling conditions, the electrical signal is randomly sampled to obtain the current reference voltage corresponding to the electrical signal. The preset sampling conditions can be preset time conditions and / or preset exposure conditions. The preset time condition is that the time difference between the current time of acquiring the electrical signal and the time of the last random sampling is not less than a preset time.

[0089] The preset exposure condition is that the current time of acquiring the electrical signal is the exposure start time. In the dynamic edge imaging method with embedded probabilistic coding provided in this embodiment of the invention, the image to be identified can be an image synthesized based on the process maps to be identified corresponding to multiple exposures.

[0090] When the current reference voltage is a voltage obtained based on preset exposure conditions, the step of generating the image to be identified based on the pulse signal and the pre-obtained pixel coordinates may include:

[0091] For each current reference voltage, multiple process images to be identified are generated based on the pulse signal corresponding to the current reference voltage and the pre-acquired pixel coordinates; the multiple process images to be identified are then synthesized to obtain the image to be identified.

[0092] For example, there are N exposures, specifically the first exposure, the second exposure, ... the Nth exposure. The reference voltage corresponding to the first exposure is the current reference voltage 1, and the image to be identified based on the current reference voltage 1 is the process to be identified. Figure 1 The reference voltage corresponding to the second exposure is the current reference voltage 2, and the image to be identified based on the current reference voltage 2 is the identification process. Figure 2 The reference voltage corresponding to the Nth exposure is the current reference voltage N. The image to be identified based on the current reference voltage N is the image to be identified process N. Figure 1 The process to be identified Figure 2 The process diagram N to be identified can be synthesized into the image a to be identified.

[0093] In order to accelerate the generation speed of the to-be-identified image, one current reference voltage can correspond to multiple exposures. For example, there are 8 exposures, specifically, a first exposure, a second exposure, …, an eighth exposure, the reference voltages corresponding to the first exposure to the fourth exposure are the current reference voltage 11, and the to-be-identified images obtained based on the current reference voltage 11 are the to-be-identified process image 11, the to-be-identified process image 12, the to-be-identified process image 13, and the to-be-identified process image 14. The reference voltages corresponding to the fifth exposure to the eighth exposure are the current reference voltage 12, and the to-be-identified images obtained based on the current reference voltage 12 are the to-be-identified process image 15, the to-be-identified process image 16, the to-be-identified process image 17, and the to-be-identified process image 18. Further, the to-be-identified image b can be synthesized based on the to-be-identified process image 11, the to-be-identified process image 12, …, the to-be-identified process image 18.

[0094] In the embodiment, the electrical signal can be randomly sampled after each preset time period, and the corresponding current reference voltage can be obtained, so that the frame rate of the image sensor can be improved. The electrical signal can also be randomly sampled after each exposure is performed, so that the highest dynamic response can be obtained. The actual use case can be used for setting.

[0095] In the case of a preset sampling condition being a preset time condition, that is, in the case that the time difference between the current time of collecting the electrical signal and the time of the last random sampling is not less than a preset time period, the electrical signal can be randomly sampled to obtain the current reference voltage corresponding to the electrical signal. The preset time period can be set according to actual use requirements, for example, the preset time period can be 1s, 50ms, 20ms, which are all reasonable and are not limited here.

[0096] That is, random sampling needs to be performed every preset time period to obtain the current reference voltage. In order to facilitate understanding, the following will introduce the corresponding relationship between the current reference voltage and the electrical signal by combining specific examples.

[0097] For example, the collection time of the electrical signal A is time point A, the electrical signal A is randomly sampled to obtain the current reference voltage A corresponding to the electrical signal A. The electrical signal is collected in real time until the time period between the current time (time point B) and time point A meets the preset time period, the electrical signal B corresponding to time point B is randomly sampled to obtain the current reference voltage B corresponding to the electrical signal B. Between the electrical signal A and the electrical signal B, there are multiple time points corresponding to the electrical signal, and the multiple time points corresponding to the electrical signal have a corresponding relationship with the current reference voltage A. That is, one current reference voltage corresponds to multiple time points of the electrical signal.

[0098] Since the electrical signal corresponding to the image to be processed is not the same for each time point in the exposure process, in the embodiment, the electrical signal can be randomly sampled when the electrical signal meets the preset time condition, and the current reference voltage corresponding to the electrical signal is obtained, which is equivalent to updating the current reference voltage, so that the current reference voltage is more accurate. In this way, in the process of generating the pulsed image data (image to be recognized), the pulsed signal can be regulated by the more accurate current reference voltage, so that more accurate pulsed image data can be obtained.

[0099] As an embodiment of the present application, the step of randomly sampling the electrical signal to obtain the current reference voltage corresponding to the electrical signal can include:

[0100] Randomly sample a preset number of electrical signals from the electrical signal as sampled electrical signals.

[0101] In order to enable the current reference voltage to represent the global exposure characteristics of the electrical signal, a preset number of electrical signals can be randomly sampled as sampled electrical signals in the process of randomly sampling the electrical signal.

[0102] The preset number can be determined based on the number of pixel units. The value corresponding to the preset number is not less than 1% of the data corresponding to the pixel unit. In other words, the number of pixel units corresponding to the sampled electrical signal is not less than 1% of the total number of pixel units.

[0103] For example, the size of the pixel array is 1024x1024, and for each row / column of the pixel array, 10 electrical signals corresponding to the pixel units in the row / column are randomly selected as sampled electrical signals.

[0104] Determine the current reference voltage based on the intensity value of the sampled electrical signal.

[0105] After obtaining the sampled electrical signal, the current reference voltage can be determined based on the intensity value of the sampled electrical signal. In one embodiment, the intensity value of the sampled electrical signal can be counted, and the median value of the intensity value of the sampled electrical signal is obtained as the current reference voltage. In another embodiment, the intensity value of the sampled electrical signal can be counted, and the average value of the intensity value of the sampled electrical signal is obtained as the current reference voltage.

[0106] As an embodiment, a preset calibration ratio can be obtained. After obtaining the median value of the intensity value of the sampled electrical signal or the average value of the intensity value of the sampled electrical signal, the median value of the intensity value of the sampled electrical signal can be amplified according to the calibration ratio, or the average value of the intensity value of the sampled electrical signal can be amplified according to the calibration ratio, so as to obtain the current reference voltage.

[0107] It can be seen that in the embodiment, the current reference voltage capable of representing the global exposure characteristics of the electrical signal can be obtained, which facilitates subsequent regulation of the pulse signal by the more accurate current reference voltage, so that more accurate pulsed image data can be obtained. In addition, in the embodiment, the current reference voltage is obtained by dynamically using the probabilistic sampling and statistical method, thereby improving the dynamic range of the image.

[0108] As an embodiment of the present application, the electrical signal is the electrical signal corresponding to a plurality of pixel units. For example, the electrical signal can be the electrical signal corresponding to a row of pixel units in the pixel array, and the electrical signal can also be the electrical signal corresponding to a column of pixel units in the pixel array.

[0109] The step of gating and amplifying the electrical signal and obtaining the current voltage based on the current gated and amplified electrical signal can include:

[0110] According to a preset selection manner, the corresponding electrical signal is selected from the electrical signal for gating and amplification to obtain a voltage to be processed. The preset selection manner includes a row-by-row selection manner and a column-by-column selection manner.

[0111] That is, after the electrical signal is collected in real time, the corresponding electrical signal can be selected from the electrical signal for gating and amplification according to the row-by-row selection manner to obtain the voltage to be processed, and the voltage to be processed is the voltage corresponding to a row of pixel units in the pixel array.

[0112] After the electrical signal is collected in real time, the corresponding electrical signal can be selected from the electrical signal for gating and amplification according to the column-by-column selection manner to obtain the voltage to be processed, and the voltage to be processed is the voltage corresponding to a column of pixel units in the pixel array.

[0113] Among them, the corresponding voltage to be processed can be obtained according to the row-by-row selection manner first, and then the corresponding voltage to be processed can be obtained according to the column-by-column selection manner. The corresponding voltage to be processed can also be obtained according to the column-by-column selection manner first, and then the corresponding voltage to be processed can be obtained according to the row-by-row selection manner. This is reasonable.

[0114] The voltage corresponding to each pixel unit included in the voltage to be processed is sequentially taken as the current voltage.

[0115] After obtaining the voltage to be processed, that is, after obtaining the voltage corresponding to a column of pixel units in the pixel array or the voltage corresponding to a row of pixel units in the pixel array, the voltage corresponding to each pixel unit included in the voltage to be processed is sequentially taken as the current voltage.

[0116] That is, the current voltage is a voltage corresponding to a single pixel unit, and the current voltage includes multiple current voltages, and each current voltage corresponds to different pixel units, that is, for a row of pixel units in the pixel array, pixel units corresponding to different pixel coordinates correspond to different current voltages.

[0117] It can be seen that in the embodiment, the current voltage can be obtained to obtain the pulsed image data subsequently.

[0118] As an implementation of the embodiment of the present application, the step of generating the to-be-recognized image based on the pulse signal and the pixel coordinates obtained in advance can include:

[0119] Converting the pulse signal into a digital signal, adding the pixel coordinates corresponding to the digital signal to the digital signal, and generating the to-be-recognized image based on the digital signal to which the pixel coordinates are added.

[0120] In the case of a high-level signal, the high-level signal can be represented by a digital 1, and in the case of a low-level signal, the low-level signal can be represented by a digital 0, so as to realize the conversion of the pulse signal into digital information.

[0121] Since the digital signal has a corresponding pixel unit, the pixel coordinates of the pixel unit corresponding to the digital signal can be obtained, and for each digital signal, the pixel coordinates corresponding to the digital signal can be added to the digital signal.

[0122] In an implementation, the pixel coordinates corresponding to the digital signal can be added to the digital signal, so as to realize the addition of the pixel coordinates corresponding to the digital signal to the digital signal.

[0123] Further, after the pixel coordinates corresponding to the digital signal are added to the digital signal, the to-be-recognized image can be generated based on the digital signal to which the pixel coordinates are added.

[0124] As an implementation of the embodiment of the present application, the step of generating the to-be-recognized image based on the digital signal to which the pixel coordinates are added can include:

[0125] The digital signal to which the pixel coordinates are added is sorted to generate the to-be-recognized image.

[0126] After the digital signal to which the pixel coordinates are added is obtained, the digital signal can be sorted according to the positions corresponding to the pixel coordinates, so that the to-be-recognized image can be generated. In an implementation, the pixel signal corresponding to each row of pixel units or the pixel signal corresponding to each column of pixel units can be sorted, so that the to-be-recognized image can be generated.

[0127] As one embodiment of the present invention, after the step of generating the image to be recognized described above, the method may further include:

[0128] The image to be identified is input into a pre-trained spiking neural network for image recognition, and the recognition result is obtained.

[0129] In one implementation, after obtaining the digital signal with added pixel coordinates, the pixel signals corresponding to each row of pixel units or each column of pixel units can be sorted, so that the pixel signals are transmitted in parallel to the spiking neural network in row order or column order, and the spiking neural network can then perform image recognition and obtain the image recognition result.

[0130] To better understand the dynamic edge imaging method with embedded probabilistic coding provided in this embodiment of the invention, the following is combined with... Figure 2 and Figure 3 This invention introduces a dynamic edge imaging method with embedded probabilistic coding provided by an embodiment of the present invention:

[0131] like Figure 2 As shown, Figure 2 The image to be processed is an example of the original image processed using the dynamic edge imaging method with embedded probabilistic coding provided in this embodiment of the invention.

[0132] in, Figure 2 The image is a grayscale image of a car in motion, using 256 levels of grayscale, and the image size is 900 pixels x 700 pixels.

[0133] like Figure 3 As shown, Figure 3 The image to be identified is an example of an image processed using a dynamic edge imaging method with embedded probabilistic coding provided in this embodiment of the invention.

[0134] according to Figure 3 It is possible to distinguish the main content of the car's driving scene, and Figure 3 The corresponding image size is only Figure 2 12.5%. In other words, after processing the image using the dynamic edge imaging method with embedded probabilistic coding provided in this embodiment of the invention, image recognition can still be guaranteed, and the recognition speed of the spiking neural network can be improved, thereby enhancing the user experience.

[0135] The following describes the dynamic edge imaging device with embedded probabilistic coding provided by the present invention. The dynamic edge imaging device with embedded probabilistic coding described below and the dynamic edge imaging method with embedded probabilistic coding described above can be referred to in correspondence.

[0136] likeFigure 4 As shown in the embodiments of the present application, a dynamic edge imaging device with embedded probability coding is provided, which comprises:

[0137] The image acquisition unit 410 is configured to acquire an electrical signal corresponding to the to-be-processed image in real time when the to-be-processed image is collected.

[0138] The gating and amplifying unit 420 is configured to gate and amplify the electrical signal, and obtain a current voltage based on the current gated and amplified electrical signal, wherein the current voltage is a voltage corresponding to a single pixel unit, and the current voltage includes a plurality of different corresponding pixel units.

[0139] The pulse signal acquisition unit 430 is configured to acquire a pulse signal based on a predetermined current reference voltage, a predetermined to-be-compared voltage, and a relationship between the current voltage, wherein the current reference voltage has a corresponding relationship with the current voltage.

[0140] The generation unit 440 is configured to generate a to-be-recognized image based on the pulse signal and a predetermined pixel coordinate.

[0141] As an embodiment of the present application, as shown in the embodiments of the present application, Figure 5 The pulse signal acquisition unit 430 can be a random circuit, which can include a transistor 510, a magnetic tunnel junction 520, a comparator 530, and a power supply (not labeled in the figure).

[0142] The current voltage is an input voltage of a gate of the transistor 510, a drain of the transistor 510 is connected to one end of the magnetic tunnel junction 520, the other end of the magnetic tunnel junction 520 is connected to a negative electrode (-VDD1) of the power supply, a positive electrode of the power supply (VDD1) is connected to a source of the transistor 510, the current reference voltage is an input voltage of a positive input end of the comparator 530, a negative input end of the comparator 530 is connected to the source of the transistor 510, and the negative input end of the comparator 530 is connected to a positive electrode of the power supply.

[0143] The to-be-compared voltage includes a driving voltage of the transistor 510 and a voltage division voltage, and the voltage division voltage is a voltage obtained based on a supply voltage and a voltage of the magnetic tunnel junction 520, wherein the supply voltage is a voltage provided by the power supply.

[0144] In the embodiments of the present application, Figure 5 The positive electrode of the power supply is connected to the source of the transistor 510 through a matching resistor. One end of the matching resistor is connected to the positive electrode of the power supply, the other end of the matching resistor is connected to the source of the transistor 510, and the negative input end of the comparator 530 is connected to the other end of the matching resistor.

[0145] The matching resistor can function as a voltage divider together with the magnetic tunnel junction 520, that is, the divided voltage can be a voltage obtained based on the supply voltage, the voltage of the magnetic tunnel junction 520, and the voltage of the matching resistor. As an implementation, the matching resistor can be an adjustable resistor, and the non-uniformity introduced by the magnetic tunnel junction 520 in the manufacturing process can be eliminated by adjusting the resistance value of the matching resistor.

[0146] The pulse signal is output by the comparator 530 based on the relationship among the current reference voltage, the voltage to be compared, and the current voltage.

[0147] The logic of the stochastic circuit can be that whether the transistor 510 is turned on is determined based on the characteristics of the transistor 510, the current voltage, and the driving voltage. The transistor 510 functions as a switch in the stochastic circuit, and specifically, the transistor 510 can be determined to be turned on or not turned on according to the current voltage input to the gate of the transistor 510.

[0148] In the case where the transistor 510 is turned on, the voltage input to the negative input terminal of the comparator 530 is the divided voltage, and the voltage input to the positive input terminal of the comparator 530 is the current reference voltage. Therefore, the comparator 530 can compare the divided voltage and the current reference voltage.

[0149] In the case where the divided voltage is greater than the current reference voltage, the comparator 530 outputs a high-level signal, and in the case where the divided voltage is not greater than the current reference voltage, the comparator 530 outputs a low-level signal.

[0150] In the case where the transistor 510 is not turned on, the voltage input to the negative input terminal of the comparator 530 is the supply voltage, and the voltage input to the positive input terminal of the comparator 530 is the current reference voltage. Therefore, the comparator 530 can compare the supply voltage and the current reference voltage.

[0151] In the case where the supply voltage is greater than the current reference voltage, the comparator 530 outputs a high-level signal, and in the case where the supply voltage is not greater than the current reference voltage, the comparator 530 outputs a low-level signal.

[0152] The regulation of the stochastic circuit is embodied in that, in the case where the current voltage input to the gate of the transistor 510 has a greater negative deviation from the current reference voltage, the stochastic circuit outputs a low-level signal with a greater probability, and in the case where the current voltage input to the gate of the transistor 510 has a greater positive deviation from the current reference voltage, the stochastic circuit outputs a high-level signal with a greater probability.

[0153] In an implementation, the regulation of the stochastic circuit can be achieved in that, in the case where the current voltage input to the gate of the transistor 510 is equal to the current reference voltage, the probability of the stochastic circuit outputting a high-level signal is 50%.

[0154] The random circuit can still start up by the magnetic tunnel junction 520 without external triggering, thereby completely solving the problem of data misplacement caused by mismatching of the trigger signal.

[0155] In the random circuit, the magnetic tunnel junction 520 is a non-stable device composed of two nanometer magnets, which can be compared with the current voltage input to the gate of the transistor 510, so that the random circuit can output a probabilistic pulse signal, which can be as shown in Figure 6 .

[0156] In Figure 6 , when the input voltage of the random circuit is small, that is, when the current voltage input to the gate of the transistor 510 is small, the average value of the probabilistic pulse signal output by the random circuit is 0.22 Figure 6 , as shown in (a). When the current voltage input to the gate of the transistor 510 gradually increases, the average value of the probabilistic pulse signal output by the random circuit can rise from 0.22 to 0.49 Figure 6 , as shown in (b). When the current voltage input to the gate of the transistor 510 continues to increase, the average value of the probabilistic pulse signal output by the random circuit can rise from 0.49 to 0.75 Figure 6 , as shown in (c).

[0157] , the corresponding values are the average values of the probabilistic pulse signal in 100 μs sampling time, Figure 6 , the vertical coordinate is the normalized value. Figure 6 Figure 6 , the corresponding values are the average values of the probabilistic pulse signal in 100 μs sampling time,

[0158] As an embodiment of the present application, the image acquisition unit 410 can include a color filter and an image acquisition device, wherein the color filter corresponds to the color filtering arrangement device and can be used to filter the visible light (ambient light) into monochromatic light.

[0159] , the image acquisition device can be an image sensor, since the CMOS type image sensor can only respond to light intensity and cannot respond to color, the color filter can be set so that the corresponding electrical signal of the pixel array of the image sensor has a positive correlation with the intensity of the monochromatic light, so as to subsequently obtain the to-be-recognized image.

[0160] The image sensor can acquire the signal of the to-be-processed image, and the corresponding pixel array of the image sensor can generate a corresponding electrical signal under light excitation. The image acquisition unit 410 specifically filters the visible light into monochromatic light and enters the pixel unit of the corresponding pixel array of the image sensor, and then photoelectric effect can occur to generate a corresponding electrical signal.​

[0161] In the case that the image sensor is a COMS image sensor, the COMS image sensor has the ability to distinguish each pixel unit, and the COMS image sensor can realize one-to-one correspondence between the electrical signal corresponding to each pixel unit and the pixel coordinate.

[0162] As an embodiment of the present application, before the above-mentioned pulse signal is obtained based on the relationship between the predetermined current reference voltage, the pre-acquired voltage to be compared and the current voltage, the above-mentioned device can further comprise a sampling and averaging unit. The sampling and averaging unit can be a sampler and averager.

[0163] The sampling and averaging unit is configured to randomly sample the electrical signal to obtain a current reference voltage corresponding to the electrical signal when the electrical signal meets a preset sampling condition.

[0164] As an embodiment of the present application, the sampling and averaging unit is specifically configured to randomly sample a preset number of electrical signals from the electrical signal as sampled electrical signals, and determine the current reference voltage based on the intensity values of the sampled electrical signals.

[0165] As an embodiment of the present application, the electrical signal is an electrical signal corresponding to a plurality of pixel units.

[0166] The above-mentioned gating and amplifying unit 420 can comprise a gate and an amplifier. The number of channels of the amplifier is consistent with the arrangement of the pixel array.

[0167] The gate and the amplifier select and amplify the corresponding electrical signal from the electrical signal according to a preset selection mode to obtain a voltage to be processed.

[0168] The preset selection mode includes a row-by-row selection mode and a column-by-column selection mode.

[0169] The voltage corresponding to each pixel unit included in the voltage to be processed is sequentially taken as a current voltage.

[0170] As an embodiment of the present application, the above-mentioned generating unit 440 can comprise a buffer and a calculation subunit. For example, the calculation subunit can be a field programmable gate array (FPGA). The FPGA can be deployed with a pre-trained pulse neural network.

[0171] The buffer and the calculation subunit convert the pulse signal into a digital signal, add a pixel coordinate corresponding to the digital signal to the digital signal, and generate an image to be recognized based on the digital signal with the added pixel coordinate.

[0172] For example, in 40 clock cycles, a pulse signal is converted into a digital signal, with a digital 0 representing a low-level signal and a digital 1 representing a high-level signal; a pixel coordinate corresponding to the digital signal is added in front of the digital signal, and the row / column coordinates each occupy 12 bits, forming a 64-bit data block. In this way, the row / column coordinates of the pixel unit each occupy 12 bits, which can be compatible with most CMOS sensors at present.

[0173] As an implementation form of the embodiment of the present application, the calculation subunit is specifically configured to sort the digital signal after the pixel coordinate is added, and generate the to-be-recognized image.

[0174] As an implementation form of the embodiment of the present application, the calculation subunit is further configured to input the to-be-recognized image to a pre-trained pulse neural network for image recognition after the to-be-recognized image is generated, and obtain a recognition result.

[0175] As an implementation form of the embodiment of the present application, the dynamic edge imaging device with embedded probability coding can further include a timing circuit unit and a position coding unit.

[0176] The timing circuit unit is configured to provide a timing signal to the position coding unit, the image acquisition device and the buffer, and the position coding unit is configured to provide a trigger signal to the gate and the buffer. The position coding unit can encode the row / column information of the pixel array into the trigger signal based on the timing signal sent by the timing circuit unit.

[0177] For example, the image sensor includes a pixel array of DxL, where D is the number of rows and L is the number of columns, and both D and L are positive integers. The trigger frequency of the image sensor is f, and the trigger frequency of the position coding unit is not greater than Dxf (corresponding to a row-by-row gating mode), or the trigger frequency of the position coding unit is not greater than Lxf (corresponding to a column-by-column gating mode). The trigger frequency of the buffer is not less than DxLxf.

[0178] In an implementation form, the position coding unit can encode the row / column information of the pixel array as the number of high-level signals or the number of low-level signals. Correspondingly, the gate can count the high-level signals or the low-level signals, so that the parameter information corresponding to the electrical signal of the pixel unit to be gated and amplified can be obtained, where the parameter information can correspond to the number of high-level signals or the number of low-level signals.

[0179] Next, an embedded probability coding dynamic edge imaging device provided by an embodiment of the present application is described by taking Figure 7 as an example.

[0180] For example, the image sensor includes a pixel array of DxL, where D is the number of rows and L is the number of columns, and both D and L are positive integers. The trigger frequency of the image sensor is f, and the trigger frequency of the position coding unit is not greater than Dxf (corresponding to a row-by-row gating mode), or the trigger frequency of the position coding unit is not greater than Lxf (corresponding to a column-by-column gating mode). The trigger frequency of the buffer is not less than DxLxf. Figure 7As shown, the dynamic edge imaging device with embedded probability coding provided by the embodiment of the present application comprises a color filter 701, a pixel sensor 702, a gate 703, a sampler and averager 704, an amplifier 705, a position encoder 706, a timing circuit 707, a random circuit 708, a buffer 709 and a calculation module 710.

[0181] The color filter 701 is a color filter arrangement, the pixel sensor 702 is an image sensor (image acquisition device), and the sampler and averager 704 is a sampler and averager unit, and the timing circuit 707 is a timing circuit unit.

[0182] The color filter 701 is connected with the pixel sensor 702, the pixel sensor 702 is connected with the timing circuit 707, the pixel sensor 702 is also connected with the sampler and averager 704, and the pixel sensor 702 is also connected with the gate 703.

[0183] The gate 703 is connected with the amplifier 705, and the amplifier 705 is connected with the gate of the transistor in the random circuit 708. The gate 703 is also connected with the position encoder 706, the position encoder 706 is connected with the timing circuit 707, and the position encoder 706 is also connected with the buffer 709.

[0184] The timing circuit 707 is also connected with the buffer 709, the buffer 709 is connected with the calculation module 710, the buffer 709 is also connected with the output end of the comparator in the random circuit 708, and the output end of the comparator in the random circuit 708 is connected with the sampler and averager 704.

[0185] It can be seen that the dynamic edge imaging device with embedded probability coding provided by the present application can integrate the acquisition of the image to be processed and the pulse conversion process of the image to be processed into the dynamic edge imaging device with embedded probability coding, and can complete the image recognition in the dynamic edge imaging device with embedded probability coding.

[0186] That is, the dynamic edge imaging device with embedded probability coding provided by the present application can adopt a sensing, coding and calculation edge combination scheme to perform synchronous processing of the whole row / column pixel information in a parallel manner, and can realize real-time and rapid recognition of the image to be recognized.

[0187] The dynamic edge imaging device with embedded probability coding provided by the present application can acquire the pulsed image data, realizes the acquisition of the pulsed image data in the process of collecting the image to be processed, solves the defect of delay in the pulsed processing of the image data, and can improve the use experience of the user. For example, the dynamic edge imaging device with embedded probability coding provided by the present application can be applied to the scene requiring real-time recognition such as automatic driving.

[0188] It should be pointed out finally that the above embodiments are only used to illustrate the technical solutions of the present application, but not to limit the same; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that the technical solutions recorded in the foregoing embodiments can still be modified, or some technical features therein can be replaced equivalently; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method of dynamic edge imaging with in-line probabilistic encoding, characterized by, The method comprises: In the case of collecting an image to be processed, the corresponding electrical signal of the image to be processed is acquired in real time; The electrical signal is gated and amplified, and based on the current gated and amplified electrical signal, a current voltage is obtained, wherein the current voltage is the voltage corresponding to a single pixel unit, and the current voltage includes multiple and corresponding pixel units which are different; Based on the relationship between the predetermined current reference voltage, the pre-acquired comparison voltage and the current voltage, a pulse signal is acquired, wherein the current reference voltage and the current voltage have a corresponding relationship; Based on the pulse signal and the pre-acquired pixel coordinates, a to-be-recognized image is generated; The comparison voltage includes the driving voltage of the transistor and the voltage divider voltage, and the voltage divider voltage is the voltage acquired based on the supply voltage and the magnetic tunnel junction voltage; the step of acquiring the pulse signal based on the relationship between the predetermined current reference voltage, the pre-acquired comparison voltage and the current voltage comprises: determining whether the transistor is turned on based on the characteristics of the transistor, the current voltage and the driving voltage; in the case that the transistor is turned on, comparing the voltage divider voltage and the current reference voltage; in the case that the voltage divider voltage is greater than the current reference voltage, outputting a high-level signal; in the case that the voltage divider voltage is not greater than the current reference voltage, outputting a low-level signal; in the case that the transistor is not turned on, comparing the supply voltage and the current reference voltage; in the case that the supply voltage is greater than the current reference voltage, outputting a high-level signal; in the case that the supply voltage is not greater than the current reference voltage, outputting a low-level signal.

2. The in-lens probabilistic encoding dynamic edge imaging method of claim 1, wherein, Before the step of acquiring the pulse signal based on the relationship between the predetermined current reference voltage, the pre-acquired comparison voltage and the current voltage, the method further comprises: In the case that the electrical signal meets the preset sampling condition, the electrical signal is randomly sampled to obtain the current reference voltage corresponding to the electrical signal.

3. The in-lens probabilistic encoding dynamic edge imaging method of claim 2, wherein, The step of randomly sampling the electrical signal to obtain the current reference voltage corresponding to the electrical signal comprises: Randomly sampling a preset number of electrical signals from the electrical signal as sampling electrical signals; Based on the intensity value of the sampling electrical signal, the current reference voltage is determined.

4. The in-lens probabilistic encoding dynamic edge imaging method of claim 1, wherein, The electrical signal is the electrical signal corresponding to a plurality of pixel units; The step of gating and amplifying the electrical signal to obtain the current voltage comprises: According to a preset selection method, the corresponding electrical signal is selected from the electrical signal for gating and amplification to obtain a to-be-processed voltage, wherein the preset selection method includes a row-by-row selection method and a column-by-column selection method; The voltage corresponding to each pixel unit included in the to-be-processed voltage is sequentially taken as the current voltage.

5. The in-lens probabilistic encoding dynamic edge imaging method according to any one of claims 1-4, characterized in that, The step of generating a to-be-recognized image based on the pulse signal and the pre-acquired pixel coordinates comprises: The pulse signal is converted into a digital signal, pixel coordinates corresponding to the digital signal are added to the digital signal, and a to-be-recognized image is generated based on the digital signal after adding the pixel coordinates.

6. The in-lens probabilistically encoded dynamic edge imaging method of claim 5, wherein, The step of generating the to-be-recognized image based on the digital signal after adding the pixel coordinates comprises: The digital signal after adding the pixel coordinates is sorted to generate the to-be-recognized image.

7. The in-lens probabilistic encoding dynamic edge imaging method according to any one of claims 1-4, wherein, After the step of generating the to-be-recognized image, the method further comprises: The to-be-recognized image is input into a pre-trained pulse neural network for image recognition to obtain a recognition result.

8. An in-line probabilistic encoding dynamic edge imaging apparatus, characterized by, The device comprises: An image acquisition unit configured to acquire, in real time, an electrical signal corresponding to a to-be-processed image when the to-be-processed image is collected; A gating and amplification unit configured to gate and amplify the electrical signal and obtain a current voltage based on the electrical signal after the gating and amplification, wherein the current voltage is a voltage corresponding to a single pixel unit, and the current voltage includes a plurality of different corresponding pixel units; A pulse signal acquisition unit configured to acquire a pulse signal based on a predetermined current reference voltage, a to-be-compared voltage pre-acquired, and a relationship between the current voltage, wherein the current reference voltage has a corresponding relationship with the current voltage; the to-be-compared voltage includes a driving voltage of a transistor and a voltage division voltage, and the voltage division voltage is a voltage acquired based on a supply voltage and a magnetic tunnel junction voltage; the step of acquiring the pulse signal based on the predetermined current reference voltage, the to-be-compared voltage pre-acquired, and the relationship between the current voltage comprises: determining whether the transistor is turned on based on characteristics of the transistor, the current voltage, and the driving voltage; comparing the voltage division voltage and the current reference voltage in a case where the transistor is turned on; outputting a high-level signal in a case where the voltage division voltage is greater than the current reference voltage; outputting a low-level signal in a case where the voltage division voltage is not greater than the current reference voltage; comparing the supply voltage and the current reference voltage in a case where the transistor is not turned on; outputting a high-level signal in a case where the supply voltage is greater than the current reference voltage; and outputting a low-level signal in a case where the supply voltage is not greater than the current reference voltage; A generation unit configured to generate a to-be-recognized image based on the pulse signal and a pre-acquired pixel coordinate.

9. The inline probabilistically encoded dynamic edge imaging device of claim 8, wherein, The pulse signal acquisition unit comprises a transistor, a magnetic tunnel junction, a comparator, and a power supply. The current voltage is an input voltage of a gate of the transistor, a drain of the transistor is connected to one end of the magnetic tunnel junction, the other end of the magnetic tunnel junction is connected to a negative electrode of the power supply, a positive electrode of the power supply is connected to a source of the transistor, and the current reference voltage is an input voltage of a positive input end of the comparator, and a negative input end of the comparator is connected to the source of the transistor. The to-be-compared voltage includes the driving voltage of the transistor and the voltage division voltage, and the voltage division voltage is a voltage acquired based on the supply voltage and the magnetic tunnel junction voltage, and the supply voltage is a voltage provided by the power supply. The pulse signal is outputted by the comparator based on the relationship among the current reference voltage, the voltage to be compared and the current voltage.

Images