A distributed event camera synchronization method using time coding

Abstract

The present invention discloses a method for synchronizing distributed event cameras using time coding, which belongs to the field of image acquisition and aims to achieve time synchronization of distributed event camera acquisition systems through precise time coding. The system is composed of Beidou / GPS timing equipment, event cameras and FPGA boards. Accurate time information is obtained through Beidou / GPS timing equipment as a time reference, and these time information are divided into binary codes; the encoded time information is then processed using the FPGA board to generate a nine-bit pulse sequence containing time information, and these signals are triggered and input into the event camera in a predetermined order; the event camera records the timestamp information at the time of triggering, and the pulse sequence is extracted by analysis to achieve timestamp alignment of the event camera, thereby completing time synchronization between multiple distributed event cameras.

Description

Technical Field

[0001] The invention belongs to the field of image acquisition, and in particular refers to a distributed event camera synchronization method using time coding. Background Art

[0002] As a new type of visual sensor, event cameras have unique advantages and characteristics in their asynchronous and low-latency data output compared to standard cameras that capture images at a fixed frame rate. Each pixel of an event camera is equipped with an independent photoelectric sensor module. When the brightness change of a pixel exceeds the set threshold, event data is asynchronously generated and output, and the data is sparse in space, reducing data redundancy. In addition, in dynamic scenes, event cameras can also capture rich visual information in environments with extremely poor lighting or extremely high brightness, and the low latency characteristics of event cameras enable them to effectively avoid motion blur in high-speed motion scenes. Due to the asynchronous and low-latency characteristics of event cameras and their ability to output data in microseconds, multiple event cameras are used for multi-view and high-precision data acquisition tasks in visual fields such as target detection, scene reconstruction, and posture estimation.

[0003] The timestamp sequence generated by the event camera is based on the startup time. In a distributed event camera acquisition system, multiple event cameras are distributed in different locations and cannot be started at the same time. Therefore, a common time reference is required to calibrate the timestamps to achieve timestamp alignment. At present, there are two methods for time synchronization between multiple event cameras: 1) using synchronization lines to connect and set up master and slave cameras to achieve synchronization; 2) using trigger lines to connect and use hardware triggers to achieve synchronization. However, these two methods are only suitable for short-distance connections between event cameras. For long-distance distributed application scenarios, trigger lines and synchronization lines are difficult to achieve long-distance connections, and long-distance connections will also cause significant signal attenuation, affecting synchronization accuracy.

[0004] The present invention is a method for solving the time synchronization problem of deploying a remote distributed event camera acquisition system. Summary of the invention

[0005] In order to solve the above technical problems, the present invention provides a method for synchronizing distributed event cameras using time coding, which can complete the time synchronization task after data collection in the distributed event camera system.

[0006] The present invention provides a distributed event camera synchronization method using time coding, the method comprising the following steps:

[0007] Step 1: Obtain accurate absolute time information from Beidou or GPS timing equipment; use satellite signals from GPS and Beidou navigation systems to provide nanosecond-level time accuracy;

[0008] Step 2: The FPGA board divides and processes the time information and performs binary encoding;

[0009] Step 3: Convert the binary coded information into a pulse train through the FPGA board;

[0010] Step 4: Extract and analyze the pulse train recorded by the event camera and align the timestamps.

[0011] Furthermore, the specific method of step 2 is as follows:

[0012] The FPGA board divides the time information provided by the timing device and uses a five-bit binary code to represent the time range of 0 to 23 hours. Each hour is further divided into six time periods, each of which is ten minutes long. Next, a three-bit binary code is used to identify the time periods from 0 to 5. Finally, the hour and time period codes are combined to form an eight-bit binary code to sequentially represent the complete time information.

[0013] This solves the problem that the time information provided by Beidou or GPS timing devices is too lengthy. For distributed event camera data collection tasks, the time period usually does not exceed 24 hours, and the time synchronization between event cameras requires a common time reference.

[0014] Furthermore, the specific method of step 3 is as follows:

[0015] The FPGA board first provides a high-level pulse as a pulse reference, and then uses the counter to generate a pulse signal with the processed eight-bit binary time coding information. The high or low level of the pulse output is set according to the current bit of the binary data to generate a nine-bit pulse sequence, and the time interval of the pulse sequence output can be customized. , where the time interval after the first pulse is Set greater than .

[0016] Furthermore, the step 4 analyzes the pulse signal sequence and aligns the timestamps. The specific method is as follows:

[0017] The data format of the event stream file captured by the event camera is , the recorded trigger signal and the corresponding timestamp information are , where the timestamp t has a time accuracy of microseconds, the first bit of the nine-bit pulse sequence obtained by the event camera is the pulse reference, and the time interval Greater than , and then the eight-bit time pulse sequence is arranged according to the time interval. The sequence position is determined by the recorded pulse signal and the corresponding event timestamp, the nine-bit binary sequence is decoded, the absolute time information is solved, and the corresponding timestamp is extracted, so as to align the event stream data of multiple event cameras.

[0018] The beneficial effects of the present invention are as follows: the present invention utilizes Beidou or GPS timing equipment and FPGA board hardware to synchronize event cameras, and adopts a designed time encoding method to perform time synchronization of multiple event cameras, which is suitable for long-distance distributed event camera acquisition scenarios and has high time synchronization accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 A device connection diagram for distributed event camera synchronization using time coding in the present invention;

[0020] Figure 2 It is a flow chart of the method for synchronizing distributed event cameras using time coding according to the present invention;

[0021] Figure 3 A schematic diagram of a pulse sequence sent by an FPGA board according to the present invention;

[0022] Figure 4 It is a schematic diagram of using a pulse sequence to perform timestamp alignment according to the present invention. DETAILED DESCRIPTION

[0023] The technical solution of the present invention will be further described in detail below in conjunction with the accompanying drawings and implementation examples of the present invention. Obviously, the implementation examples described are only part of the embodiments of the present invention, which are used to describe the present invention and are not used to limit the present invention.

[0024] The preferred distributed event camera time synchronization solution of the present invention is as follows: Figure 1-Figure 4 As shown, taking two groups of distributed event cameras as an example, the synchronization system includes two groups of Beidou or GPS timing devices and FPGA boards, event camera #1 and event camera #2.

[0025] Specifically, the Beidou or GPS timing device is connected to the FPGA board to provide precise time information. The FPGA board is connected to the event camera to send a synchronous pulse signal containing a time series. The host computer is used to control the FPGA board and the data transmission between the event camera.

[0026] A preferred implementation example further describing the method for distributed event camera synchronization using time coding includes:

[0027] Step 1: Obtain accurate absolute time information from Beidou or GPS timing equipment. Using satellite signals from GPS and Beidou navigation systems, accurate time at the nanosecond level is provided.

[0028] Step 2: The FPGA board divides and processes the time information and performs binary encoding.

[0029] The time information provided by Beidou or GPS timing equipment is too lengthy. For distributed event camera data collection tasks, the time period usually does not exceed 24 hours, and the time synchronization between event cameras requires a common time reference. Therefore, the FPGA board divides the time information provided by the timing equipment, using five-bit binary code to represent the time range of 0 to 23 hours; for each hour, it is further divided into six time periods, each of which is a unit of ten minutes; then, three-bit binary code is used to identify the time period from 0 to 5; finally, the hour and time period codes are combined to form an eight-bit binary code to sequentially represent the complete time information.

[0030] Step 3, convert the binary coded information into a pulse sequence through the FPGA board. The FPGA board first provides a high-level pulse as a pulse reference, and then uses a counter to generate a pulse signal for the processed eight-bit binary time coded information. The high or low level of the pulse output is set according to the current bit of the binary data. A total of nine-bit pulse sequences are generated, and the time interval of the pulse sequence output can be customized. In order to distinguish the pulse reference and the time pulse sequence, the time interval after the first pulse is Set greater than ,like Figure 3 As shown, the FPGA board outputs the processed pulse sequence to the event camera at time intervals, where the time interval of the first pulse is .

[0031] Step 4: Extract and analyze the pulse sequence recorded by the event camera, align the timestamps, and the data format of the event stream file captured by the event camera is , the recorded trigger signal and the corresponding timestamp information are , where the timestamp t has a time accuracy of microseconds. Figure 4 As shown, the event camera receives a nine-bit pulse sequence every ten minutes. The first bit in the pulse sequence is the pulse reference, and the time interval is

[0032] , and then the eight-bit time pulse sequence is timed according to the time interval Arrangement. The sequence position is determined by the recorded pulse signal and the corresponding event timestamp, the nine-bit binary sequence is decoded, the absolute time information is solved, and the corresponding timestamp is extracted to align the event stream data of event camera #1 and event camera #2.

[0033] The above description is only a preferred implementation scheme of the present invention and is not limited to the above implementation examples. Any modifications made by ordinary technicians in this technical field under the concept of the present invention shall fall within the protection scope of the present invention.

Claims

1. A distributed event camera synchronization method using time coding, characterized in that: The method comprises the following steps: Step 1: Obtain accurate absolute time information from Beidou or GPS timing equipment; use satellite signals from GPS and Beidou navigation systems to provide nanosecond-level time accuracy; Step 2: The FPGA board divides and processes the time information and performs binary encoding; Step 3: Convert the binary coded information into a pulse train through the FPGA board; Step 4: Extract and analyze the pulse sequence recorded by the event camera and align the timestamps; The specific method of step 2 is as follows: The FPGA board divides the time information provided by the timing device and uses five-bit binary code to represent the time range of 0 to 23 hours; for each hour, it is further divided into six time periods, each of which is ten minutes. Next, use three-bit binary code to identify the time period from 0 to 5; Finally, the hour and time period codes are combined to form an eight-bit binary code to sequentially represent the complete time information; The specific method of step 3 is as follows: The FPGA board first provides a high-level pulse as a pulse reference, and then uses the counter to generate a pulse signal with the processed eight-bit binary time coding information. The high or low level of the pulse output is set according to the current bit of the binary data to generate a nine-bit pulse sequence, and the time interval of the pulse sequence output can be customized. , where the time interval after the first pulse is Set greater than ; The step 4 analyzes the pulse signal sequence and aligns the timestamps. The specific method is as follows: The data format of the event stream file captured by the event camera is , the recorded trigger signal and the corresponding timestamp information are , where the timestamp t has a time accuracy of microseconds, the first bit of the nine-bit pulse sequence obtained by the event camera is the pulse reference, and the time interval Greater than , and then the eight-bit time pulse sequence is arranged according to the time interval; The sequence position is determined by the recorded pulse signal and the corresponding event timestamp, the nine-bit binary sequence is decoded, the absolute time information is solved, and the corresponding timestamp is extracted, thereby aligning the event stream data of multiple event cameras.

Images