Event stream denoising method and device

Abstract

The invention relates to the technical field of photoelectric detection and visual application, in particular to an event stream denoising method and device, and the method comprises the steps: obtaining the event stream information of weak target motion under an extreme illumination condition; constructing a two-dimensional array event stream quantity presentation graph according to the event stream information so as to accumulate event stream data of each pixel position; performing convolution operation on the event stream data by using a neighborhood filter to obtain a matrix neighborhood event; de-noising the event stream data according to the matrix neighborhood event to obtain de-noised event stream data and an aggregation degree value of the de-noised event stream data; and iteratively executing convolution and de-noising processes on the de-noised event stream data until the aggregation degree value is in a rising trend to obtain noiseless event stream data. Therefore, the problems that the asynchronous event stream output by the event camera contains a large amount of noise, the visualization quality of the event stream is affected, the sensitivity of brightness change detection caused by object movement is reduced, and information loss is caused are solved.

Description

Technical Field

[0001] The present invention relates to the field of photoelectric detection and vision application technology, and in particular to an event stream denoising method and device. Background Art

[0002] like Figure 1 As shown in the figure, traditional visual sensor technology faces challenges when dealing with high-speed motion scenes, mainly manifested in the blur effect of moving objects and pixel saturation when the brightness changes sharply. In addition, these sensors stick to a fixed frame rate output mode, which introduces the temporal discreteness of the data, further exacerbating the problem of loss of information details in fast-paced motion scenes. This limitation significantly weakens the navigation accuracy and positioning reliability of visual positioning systems in similar scenarios.

[0003] Event cameras, as a revolutionary achievement in the field of visual sensing, subvert the traditional synchronous imaging method and use advanced logarithmic light intensity sensing mechanism to capture scene dynamics. This sensor not only efficiently records changes in light intensity through its unique logarithmic photoreceptor, but also encodes these changes in logarithmic form, greatly enhancing contrast processing capabilities and sensitivity to subtle changes in light intensity.

[0004] However, there are still problems with event cameras in practical applications. Due to its own structure, event cameras are very sensitive to changes in ambient brightness, and the output asynchronous event stream contains a lot of noise BA (Background Activate) interference. BA noise may come from pulse noise during digital signal transmission and Gaussian noise caused by photodiodes, which has a great impact on further applications and the quality of event stream visualization. Increasing the threshold for event cameras to detect brightness changes helps reduce noise, but it reduces the sensitivity of detecting brightness changes caused by object motion, resulting in information loss. Therefore, how to solve the noise generated during the use of event cameras is an urgent problem to be solved. In other words, denoising the event stream is a very important preprocessing step before using the event stream information. Denoising the event stream information before using it will greatly reduce the difficulty of subsequent algorithm design. Summary of the invention

[0005] The present invention provides an event stream denoising method and device to solve the problem that the asynchronous event stream output by an event camera contains a large amount of noise interference, which affects the visualization quality of the event stream, reduces the sensitivity of detecting brightness changes caused by object motion, and causes information loss.

[0006] The first aspect of the present invention provides an event stream denoising method, comprising the following steps: obtaining event stream information of faint target motion under extreme lighting conditions; constructing a two-dimensional array event stream quantity presentation diagram based on the event stream information to accumulate event stream data at each pixel position; performing a convolution operation on the event stream data using a pre-constructed neighborhood filter to obtain matrix neighborhood events; denoising the event stream data based on the matrix neighborhood events to obtain denoised event stream data, and calculating a degree of aggregation value of the denoised event stream data; iteratively performing the convolution and denoising process on the denoised event stream data until the degree of aggregation value shows an upward trend to obtain noise-free event stream data.

[0007] Optionally, constructing a two-dimensional array event stream quantity presentation diagram according to the event stream information to accumulate event stream data at each pixel position includes:

[0008] Creating a blank canvas, and stacking the event flow information on the blank canvas using a preset time window to obtain a two-dimensional array event flow quantity presentation diagram;

[0009] The event stream data of each pixel position is obtained on the two-dimensional array event stream quantity presentation diagram.

[0010] Optionally, the height of the two-dimensional array event flow quantity presentation graph is determined according to the pixel height of the event flow information, and the width of the two-dimensional array event flow quantity presentation graph is determined according to the pixel width of the event flow information.

[0011] Optionally, denoising the event stream data according to the matrix neighborhood events to obtain denoised event stream data, and calculating a concentration degree value of the denoised event stream data, includes:

[0012] According to a preset spatiotemporal threshold, multiple events of the event stream data that are not in the matrix neighborhood events are removed to obtain the denoised event stream data;

[0013] Recording the pixel position and polarity of each event in the denoised event stream data;

[0014] The denoised event stream data is visualized according to the pixel position and polarity of each event to calculate the aggregation degree of the denoised event stream data.

[0015] Optionally, visualizing the denoised event stream data according to the pixel position and polarity of each event to calculate the aggregation degree value of the denoised event stream data includes:

[0016] Create an all-white RGB image;

[0017] Plotting the denoised event stream data on the all-white RGB image according to the pixel position and polarity of each event to obtain the visualized denoised event stream data;

[0018] The aggregation degree value of the denoised event stream data is calculated according to the visualized denoised event stream data.

[0019] Optionally, the step of drawing the denoised event stream data on the all-white RGB image according to the pixel position and polarity of each event to obtain the visual denoised event stream data comprises:

[0020] The events with polarity 1 in the denoised event stream data are set to red at the corresponding pixel positions in the all-white RGB image, and the events with polarity 0 in the denoised event stream data are set to blue at the corresponding pixel positions in the all-white RGB image, so as to obtain the visualized denoised event stream data.

[0021] The second aspect of the present invention provides an event stream denoising device, including: an acquisition module, used to acquire event stream information of faint target motion under extreme lighting conditions; an accumulation module, used to construct a two-dimensional array event stream quantity presentation diagram according to the event stream information to accumulate event stream data at each pixel position; a convolution operation module, used to perform a convolution operation on the event stream data using a pre-constructed neighborhood filter to obtain a matrix neighborhood event; a denoising module, used to denoise the event stream data according to the matrix neighborhood event to obtain denoised event stream data, and calculate the aggregation degree value of the denoised event stream data; an iterative denoising module, used to iteratively perform a convolution and denoising process on the denoised event stream data until the aggregation degree value shows an upward trend to obtain noise-free event stream data.

[0022] A third aspect of the present invention provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the event stream denoising method as described in the above embodiment.

[0023] A fourth aspect of the present invention provides a computer program product, which implements the above event stream denoising method when the computer program / instructions are executed by a processor.

[0024] A fifth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, which, when executed by a processor, implements the above event stream denoising method.

[0025] The event stream denoising method and device proposed in the embodiment of the present invention effectively filter out background activity noise through spatiotemporal correlation while retaining real event information, and show higher robustness when dealing with challenging situations (such as object boundaries and extreme lighting conditions); an event-based time context aggregation module is equipped to extract time context features in event data, and by utilizing the high time resolution characteristics of events, the network's modeling of global features is enhanced, and different event stream visualization methods are completed using different time windows; a quantitative evaluation standard is designed to evaluate the noise removal effect and image quality of the event stream, and then the denoising process can be further performed based on the noise removal effect and image quality, thereby improving the accuracy and robustness of the system noise reduction.

[0026] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0028] Figure 1 A comparison diagram of output formats of images captured by a traditional camera and an event camera provided in an embodiment of the present invention;

[0029] Figure 2 A flow chart of an event stream denoising method provided by an embodiment of the present invention;

[0030] Figure 3 A schematic diagram of an event camera provided by an embodiment of the present invention;

[0031] Figure 4 A schematic diagram of a specific implementation of the event stream denoising method provided by an embodiment of the present invention;

[0032] Figure 5 A schematic diagram of the event stream denoising principle provided by an embodiment of the present invention;

[0033] Figure 6 The event stream information visualization effect diagram provided by the embodiment of the present invention, wherein (a) is a visualization effect diagram of the event stream with the time threshold set to 10000, (b) is a visualization effect diagram of the event stream with the time threshold set to 100000, (c) is a visualization effect diagram of the denoised event stream when the time threshold is set to 10000, and (d) is a visualization effect diagram of the denoised event stream when the time threshold is set to 100000;

[0034] Figure 7A schematic diagram of the effect of denoising multiple event stream information provided by an embodiment of the present invention using an event stream denoising method;

[0035] Figure 8 A block diagram of an event stream denoising device provided by an embodiment of the present invention;

[0036] Fig. 9 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0037] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.

[0038] The event stream denoising method and device according to the embodiments of the present invention are described below with reference to the accompanying drawings.

[0039] Figure 2 A schematic diagram of a flow chart of an event stream denoising method provided by an embodiment of the present invention.

[0040] like Figure 2 As shown, the event stream denoising method includes the following steps:

[0041] In step S201, event stream information of faint target motion under extreme illumination conditions is obtained.

[0042] In the actual implementation process, you can build Figure 3 The event camera in the example uses the event camera to obtain the event flow information of the weak target motion under extreme lighting conditions, which is recorded as The noise information contained in the event stream information can be expressed as Among them, (x i ,y i ) is the pixel address where the event occurs, t i is the time of the event, p i is the polarity of the event, at p i =1 means the brightness of the event flow at this point increases, p i =0 means the brightness of the event stream at that point decreases, i=1,2L,N, where N is the total number of event stream information.

[0043] In step S202, a two-dimensional array event flow quantity presentation diagram is constructed according to the event flow information to accumulate event flow data at each pixel position.

[0044] In some embodiments, the height of the two-dimensional array event flow quantity presentation graph is determined based on the pixel height of the event flow information, and the width of the two-dimensional array event flow quantity presentation graph is determined based on the pixel width of the event flow information.

[0045] In some embodiments, a two-dimensional array event stream quantity presentation diagram is constructed according to the event stream information to accumulate event stream data at each pixel position, including:

[0046] Create a blank canvas, and use the preset time window to stack the event flow information on the blank canvas to obtain a two-dimensional array event flow quantity presentation diagram;

[0047] Get the event flow data of each pixel position on the two-dimensional array event flow quantity presentation chart.

[0048] In the actual implementation process, Figure 4 As shown, the pixel height and pixel width of the event stream information are extracted in advance to define a decorator for accelerating the event counting process, that is, a blank canvas with a dimension of (height, width) is constructed according to the pixel height and pixel width of the event stream information, and each event in the event stream information of the structured array (that is, the event stream data at each pixel position) within the preset time window is traversed to stack each event on the blank canvas to obtain a two-dimensional array event stream quantity presentation diagram, and for each event, the count value is increased at the corresponding (y, x) position of the two-dimensional array event stream quantity presentation diagram to accumulate the number of events at each pixel position in the event stream information, that is, the event stream data.

[0049] It should be noted that the preset time window can be manually set to a larger time window fixed value, and then the initial two-dimensional array event flow quantity presentation diagram can be constructed, and the observation image quality of the initial two-dimensional array event flow quantity presentation diagram can be observed. If the observation image quality is not improved, the binary search method can be used to continuously select suitable time windows in a larger time window fixed value, so that the constructed two-dimensional array event flow quantity presentation diagram has the best imaging effect.

[0050] In step S203, a convolution operation is performed on the event stream data using a pre-constructed neighborhood filter to obtain a matrix neighborhood event.

[0051] In the actual implementation process, a matrix with 3 rows and 3 columns with 0 in the middle and 1 in the rest can be pre-constructed as a neighborhood filter. The neighborhood filter is used to perform convolution operation on the event stream data to determine the number of events around the center point to be tested, so as to output a matrix neighborhood event representing the total number of events in each pixel and its neighborhood. The matrix neighborhood event can then be used to detect the edge points of the object contour, and the events with less than 3 surrounding numbers can be regarded as noise.

[0052] In step S204, the event stream data is denoised according to the matrix neighborhood events to obtain denoised event stream data, and the aggregation degree value of the denoised event stream data is calculated.

[0053] In some embodiments, denoising the event stream data according to the matrix neighborhood events to obtain the denoised event stream data, and calculating the aggregation degree value of the denoised event stream data includes:

[0054] According to a preset spatiotemporal threshold, multiple events of the event stream data that are not in the matrix neighborhood events are removed to obtain denoised event stream data;

[0055] Record the pixel position and polarity of each event in the denoised event stream data;

[0056] The denoised event stream data is visualized according to the pixel position and polarity of each event to calculate the degree of aggregation of the denoised event stream data.

[0057] like Figure 5 As shown in the figure, in the actual execution process, the edge contour points of the event stream data are detected according to the matrix neighborhood events, and the events outside the edge are removed to obtain the initial denoised event stream data. Then, the initial denoised event stream data is filtered again using the preset spatiotemporal threshold, and only those events whose neighborhood event counts exceed the threshold are retained, and their coordinates (x, y) and polarity p are recorded to obtain the denoised event stream data. The formula is:

[0058]

[0059] Among them, event fact' is the real event retained after filtering, is the horizontal coordinate of the i-th or i-th position, is the ordinate of the i-th or i-th position, and n is the number of event stream data.

[0060] Furthermore, if Figure 4 , 6 As shown in Figure 7, a full-white RGB image with the size of the height and weight of the noise-free event stream data is created. Each pixel of the image consists of three values, representing the intensity of the red, green, and blue channels, with a value range of 0-255. The denoised event stream data is plotted on the full-white RGB image according to the pixel position and polarity of each event. Specifically, the polarity p in the denoised event stream data is plotted on the full-white RGB image. i For events where the value is 1, the corresponding pixel position (x i ,y i ) is set to red; for the polarity p in the noise-free event stream data i For events where the pixel position (xi ,y i ) is set to blue, so as to obtain the visualized denoised event stream data, which can intuitively observe the expression style of the event stream information and the difference before and after processing the event stream information;

[0061] Finally, based on the visualized denoised event stream data, the following formulas are used to calculate the aggregation degree values ​​of the denoised event stream data.

[0062]

[0063] Among them, Q is the degree of aggregation, x i is the average value of the horizontal coordinates of all event flow information, y i is the average value of the ordinate of all event flow information, x n is the sum of the horizontal coordinates of all event flow information, y n is the sum of the vertical coordinates of all event stream information, and n is the number of event stream data.

[0064] In step S205, the convolution and denoising process is iteratively performed on the denoised event stream data until the aggregation degree value shows an upward trend, thereby obtaining noise-free event stream data.

[0065] In the actual execution process, a convolution operation is performed on the denoised event stream data using a pre-constructed neighborhood filter to obtain a new matrix neighborhood event; the event stream data is denoised according to the new matrix neighborhood event to obtain new denoised event stream data, and the aggregation degree value of the new denoised event stream data is calculated, and the two aggregation degree values ​​are compared to see whether they show an upward trend. If not, continue to iterate steps S203-S205 until an upward trend is shown, thereby obtaining noise-free event stream data.

[0066] The event stream denoising method proposed in the embodiment of the present invention has the following beneficial effects:

[0067] (1) Effectively filter out background activity noise caused by long-term operation of sensitive components or changes in illumination through spatiotemporal correlation, while retaining real event information, and showing higher robustness when dealing with challenging situations (such as object boundaries and extreme lighting conditions);

[0068] (2) Extracting temporal context features from event data, by utilizing the high temporal resolution of events, enhancing the network’s modeling of global features, and using different time windows to achieve different event flow visualization methods;

[0069] (3) A quantitative evaluation standard is designed to evaluate the noise removal effect and image quality of the event stream. Then, the denoising process can be further performed based on the noise removal effect and image quality, thereby improving the accuracy and robustness of the system noise reduction.

[0070] Next, the event stream denoising device proposed according to the embodiment of the present invention is described with reference to the accompanying drawings.

[0071] Figure 8 4 is a block diagram of an event stream denoising device according to an embodiment of the present invention.

[0072] like Figure 8 As shown, the event stream denoising device 80 includes: an acquisition module 801, an accumulation module 802, a convolution operation module 803, a denoising module 804 and an iterative denoising module 805.

[0073] Among them, the acquisition module 801 is used to obtain the event flow information of weak target motion under extreme lighting conditions. The accumulation module 802 is used to construct a two-dimensional array event flow quantity presentation diagram based on the event flow information to accumulate the event flow data at each pixel position. The convolution operation module 803 is used to perform a convolution operation on the event flow data using a pre-built neighborhood filter to obtain a matrix neighborhood event. The denoising module 804 is used to denoise the event flow data based on the matrix neighborhood events to obtain the denoised event flow data, and calculate the aggregation degree value of the denoised event flow data. The iterative denoising module 805 is used to iteratively perform the convolution and denoising process on the denoised event flow data until the aggregation degree value shows an upward trend to obtain noise-free event flow data.

[0074] It should be noted that the above explanation of the event stream denoising method embodiment is also applicable to the event stream denoising device of this embodiment, and will not be repeated here.

[0075] The event stream denoising device proposed in the embodiment of the present invention has the following beneficial effects:

[0076] (1) Effectively filter out background activity noise caused by long-term operation of sensitive components or changes in illumination through spatiotemporal correlation, while retaining real event information, and showing higher robustness when dealing with challenging situations (such as object boundaries and extreme lighting conditions);

[0077] (2) Extracting temporal context features from event data, by utilizing the high temporal resolution of events, enhancing the network’s modeling of global features, and using different time windows to achieve different event flow visualization methods;

[0078] (3) A quantitative evaluation standard is designed to evaluate the noise removal effect and image quality of the event stream. Then, the denoising process can be further performed based on the noise removal effect and image quality, thereby improving the accuracy and robustness of the system noise reduction.

[0079] Fig. 9 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. The electronic device may include:

[0080] A memory 901 , a processor 902 , and a computer program stored in the memory 901 and executable on the processor 902 .

[0081] When the processor 902 executes the program, the event stream denoising method provided in the above embodiment is implemented.

[0082] Furthermore, the electronic device further comprises:

[0083] The communication interface 903 is used for communication between the memory 901 and the processor 902 .

[0084] The memory 901 is used to store computer programs that can be executed on the processor 902 .

[0085] The memory 901 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0086] If the memory 901, the processor 902 and the communication interface 903 are implemented independently, the communication interface 903, the memory 901 and the processor 902 can be connected to each other through a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Fig. 9 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0087] Optionally, in a specific implementation, if the memory 901, the processor 902 and the communication interface 903 are integrated on a chip, the memory 901, the processor 902 and the communication interface 903 can communicate with each other through an internal interface.

[0088] The processor 902 may be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present invention.

[0089] An embodiment of the present invention further provides a computer program product, and when the computer program / instructions are executed by a processor, the above event stream denoising method is implemented.

[0090] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the above event stream denoising method is implemented.

[0091] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0092] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "N" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0093] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or N executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present invention includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present invention belong.

[0094] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or N wirings (electronic devices), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways as necessary and then storing it in a computer memory.

[0095] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiment, the N steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. If implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0096] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0097] In addition, each functional unit in each embodiment of the present invention may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0098] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present invention. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. An event stream denoising method, characterized in that: The following steps are involved: Obtain event stream information of faint target motion under extreme lighting conditions; Constructing a two-dimensional array event flow quantity presentation diagram according to the event flow information to accumulate event flow data at each pixel position; Performing a convolution operation on the event stream data using a pre-built neighborhood filter to obtain a matrix neighborhood event; De-noising the event stream data according to the matrix neighborhood events to obtain the de-noised event stream data, and calculating the aggregation degree value of the de-noised event stream data; The convolution and denoising process is iteratively performed on the denoised event stream data until the aggregation degree value shows an upward trend, thereby obtaining noise-free event stream data.

2. The event stream denoising method according to claim 1, characterized in that: The constructing a two-dimensional array event flow quantity presentation diagram according to the event flow information to accumulate event flow data at each pixel position includes: Creating a blank canvas, and stacking the event flow information on the blank canvas using a preset time window to obtain a two-dimensional array event flow quantity presentation diagram; The event stream data of each pixel position is obtained on the two-dimensional array event stream quantity presentation diagram.

3. The event stream denoising method according to claim 1 or 2, characterized in that: The height of the two-dimensional array event flow quantity presentation graph is determined according to the pixel height of the event flow information, and the width of the two-dimensional array event flow quantity presentation graph is determined according to the pixel width of the event flow information.

4. The event stream denoising method according to claim 1, characterized in that: Denoising the event stream data according to the matrix neighborhood events to obtain the denoised event stream data, and calculating the aggregation degree value of the denoised event stream data, includes: According to a preset spatiotemporal threshold, multiple events of the event stream data that are not in the matrix neighborhood events are removed to obtain the denoised event stream data; Recording the pixel position and polarity of each event in the denoised event stream data; The denoised event stream data is visualized according to the pixel position and polarity of each event to calculate the aggregation degree of the denoised event stream data.

5. The event stream denoising method according to claim 4, characterized in that: The visualizing the denoised event stream data according to the pixel position and polarity of each event to calculate the aggregation degree value of the denoised event stream data includes: Create an all-white RGB image; Plotting the denoised event stream data on the all-white RGB image according to the pixel position and polarity of each event to obtain the visualized denoised event stream data; The aggregation degree value of the denoised event stream data is calculated according to the visualized denoised event stream data.

6. The event stream denoising method according to claim 5, characterized in that: The step of drawing the denoised event stream data on the all-white RGB image according to the pixel position and polarity of each event to obtain the visual denoised event stream data comprises: The events with polarity 1 in the denoised event stream data are set to red at the corresponding pixel positions in the all-white RGB image, and the events with polarity 0 in the denoised event stream data are set to blue at the corresponding pixel positions in the all-white RGB image, so as to obtain the visualized denoised event stream data.

7. An event stream denoising device, characterized in that: include: The acquisition module is used to obtain event stream information of faint target motion under extreme lighting conditions; An accumulation module, used to construct a two-dimensional array event flow quantity presentation diagram according to the event flow information, so as to accumulate event flow data at each pixel position; A convolution operation module, used for performing a convolution operation on the event stream data using a pre-built neighborhood filter to obtain a matrix neighborhood event; A denoising module, used for denoising the event stream data according to the matrix neighborhood events to obtain denoised event stream data, and calculating a concentration degree value of the denoised event stream data; The iterative denoising module is used to iteratively perform convolution and denoising processes on the denoised event stream data until the aggregation degree value shows an upward trend, thereby obtaining noise-free event stream data.

8. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the event stream denoising method according to any one of claims 1 to 6.

9. A computer program product, characterized in that When the computer program / instruction is executed by a processor, the event stream denoising method according to any one of claims 1 to 6 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement the event stream denoising method according to any one of claims 1 to 6.

Images