Target encoding method and target lamp image decoding method

Abstract

The invention relates to the field of vision measurement, in particular to a target for image autonomous identification, a target coding method and an image analysis method. The target comprises a target rod, a target seat and a target lamp, one end of the target rod is adjustably installed on the target seat, and the target lamp is installed on the target rod and can display light on the front face and the rear face of the target rod. The target lamp comprises a rear scattering cover, a rear hyperboloid reflection cavity, a reflection cone, multiple lamp beads, a front hyperboloid reflection cavity and a front scattering cover, the rear hyperboloid reflection cavity and the front hyperboloid reflection cavity are in butt joint, the reflection cone is clamped and fixed in the rear hyperboloid reflection cavity and the front hyperboloid reflection cavity, the lamp beads are arranged on the front side and the rear side of the reflection cone in a scattered mode, and the front scattering cover is arranged at the front end of the front hyperboloid reflection cavity. And the rear scattering cover is arranged at the rear end of the rear hyperboloid reflection cavity. The coding device comprises a far-end host, more than one handheld terminal and more than one target coding module. And the handheld terminal is used for communicating with the remote host to perform database synchronization, communicating with the target coding module and performing addition, deletion, checking and modification on the target coding information.

Description

[0001] This application is a divisional application The filing date of the original application is: December 6, 2024 The application number of the original application is: 202411785423.9 The original application's invention is titled: Target for autonomous image identification, target encoding method, and image analysis method Technical Field The present invention relates to the field of visual measurement technology, and in particular to an equal-intensity, parallax-free, synchronous image splitting and multi-band imaging device. Background Art

[0002] When using photogrammetry technology for visual positioning, it is necessary to place external reference targets within the measurement range and match the targets on the camera image with the targets in the real space to establish the control field of the visual system. Currently, most reference targets remain relatively stationary with the camera. Through pre-identification and marking in the images captured by the camera, the identity labels of various targets are not included in the intrinsic attributes of the image. Their initialization marking is mainly achieved through manual confirmation, spatial position analysis and other methods. It is applicable to small-scale scenes, static cameras, and a small number of targets. However, for large-scale cluster cameras and multi-target scenes (such as Figure 1 As shown in the figure, the cameras are in dynamic cluster operation, and it is difficult to initialize the targets of each camera one by one. The number of targets recognized by each camera is large and difficult to distinguish (such as Figure 2 As shown in the figure, it is difficult to identify the target when any camera position is dynamically switched (e.g. Figure 3 In summary, it is necessary to develop a target and autonomous identification technology that can adapt to large scenes and meet the recognition requirements of cluster dynamic cameras, and embed the target identity information into the image attributes to improve the accuracy and efficiency of visual positioning.

[0003] To achieve the above vision, it is necessary to place several targets with known coordinates in space. By identifying the targets on the image and assigning the corresponding known coordinates, the camera's external elements (referring to the camera's three spatial coordinates and three rotational postures in space) can be determined. However, there is more than one target in the image, so the targets need to be numbered, and then the targets on each image need to be identified, and the identified targets need to be matched one-to-one with the numbers.

[0004] Current computer graphics technology can identify targets from background images, but assigning the identified targets to their corresponding numbers (that is, initializing the target coordinates) is difficult, and this is often done manually. This method is less efficient when the target spans a large number of images. When real-time computation is required, camera images are often binary, with little background reference, making it even more difficult to assign target numbers.

[0005] To ensure that the target's identity tag is inherently included in the image, the camera must simultaneously capture light sources of different wavelengths emanating from the same spatial location at the instant of capture, generating multiple parallax-free images. The target number information can then be obtained through image superposition. This requires the camera to be able to capture images of the same target at the same time, at different wavelengths, with equal intensity and without parallax. Optical filters can be used to capture images of different wavelengths, a process currently well-established. However, for capturing images of the same object at the same time, with equal intensity and without parallax, the main methods currently used are as follows: Method 1: Using a multi-lens multi-filter camera This method can simultaneously capture light sources of different wavelengths emitted by the same target. However, since multiple cameras are used to capture the scene at the same time, the spatial position of each camera cannot be completely consistent (e.g. Figure 4 As shown in the figure), the parallax of the captured image cannot be eliminated (as shown in the figure). Figure 5 As shown in the figure), after the multi-band images are superimposed, the images of the target cannot overlap, making it difficult to achieve self-identification of the target.

[0006] Method 2: Single-lens multi-filter camera This method can capture light sources of different wavelengths emitted by the same target at different times (such as Figure 6 As shown in the figure). Since the filter switching takes time, the camera cannot form an image until the filter switching is complete. Therefore, during the process of switching filters several times, the target or the camera may have been slightly disturbed. Therefore, the images of various wavelengths captured do not record the position of the target at the same time. Because the disturbance of the camera or the target will cause the target image to be inconsistent, it is also difficult to achieve self-identification of the target (such as Figure 7 shown).

[0007] Method 3: Optical image splitting device The key to achieving the imaging function of the aforementioned target lies in the image splitting, which is currently mainly done by optical methods. It uses a spectroscope to decompose the same light beam and then form an image, but there are the following limitations: First, the current spectroscope image splitting is only a simple image splitting, and does not have the multi-band light source filtering function, making it difficult to achieve synchronous imaging of co-located multi-wave light sources. Second, no device or method has been proposed to strictly control the parallax of the image, making it difficult to ensure the synchronous focusing and equal size of multiple images, resulting in parallax or defocus between images. Third, when the position of the camera and the light source is not specific (generally a horizontal light beam in the laboratory, and multiple and divergent light sources in reality), the incident angle of each beam of light after splitting by the spectroscope is different. After multiple refractions and reflections, it is impossible to ensure that the light intensity of each split image remains consistent, resulting in overexposure or underexposure of the image. Due to the above three limitations, the generated image has problems such as missing details and parallax (such as Figure 8As shown in the figure, both of these will result in the inability to overlap the images when the multi-band images are superimposed, making it difficult to achieve self-identification of the target. Summary of the Invention

[0008] The present invention provides an equal-intensity, parallax-free, synchronous image splitting and multi-band imaging device, which is used to solve the technical problem of the difficulty in accurately capturing target images in existing visual measurement technologies.

[0009] To achieve the above-mentioned objectives, the present invention provides an equal-intensity, parallax-free, synchronous image splitting and multi-band imaging device, comprising a lens and a frame, wherein the lens is mounted on the frame, and a spectroscopic channel and multiple imaging components are provided in the frame. The spectroscopic channel transmits equal-intensity light beams from the light incident from the lens through multi-level spectroscopic splitting and reflection to each imaging component, and the imaging component comprises a focusing lens, an imaging sensor and a band filter for allowing a light beam of a specific wavelength to pass through.

[0010] As a further improvement of the above technical solution: The imaging components are provided with four groups, and the spectroscopic channel is provided with a primary spectrometer and two secondary spectrometers. After the incident light is dispersed into two beams by the primary spectrometer, the two beams are respectively dispersed into two beams again by a secondary spectrometer, forming a total of four beams of light that are transmitted to the four groups of imaging components respectively.

[0011] The incident light is dispersed into a first beam of light and a second beam of light by a first-level beam splitter. The first beam of light is reflected by a first reflector and then directed to a second-level beam splitter. After being dispersed into two beams of light by the second-level beam splitter, the two beams are respectively reflected by a second reflector and a third reflector into a corresponding imaging component. The second beam of light is directly directed to another second-level beam splitter. After being dispersed into two beams of light by the second-level beam splitter, one beam is reflected by a fourth reflector and a fifth reflector in sequence into a corresponding imaging component, and the other beam is reflected by a sixth reflector and a seventh reflector in sequence into a corresponding imaging component.

[0012] A secondary beam splitter and an afterglow absorption cavity are provided between the secondary beam splitter and the imaging assembly.

[0013] The primary beam splitter, the secondary beam splitter and the secondary beam splitter are arranged in parallel.

[0014] The first reflector, the fourth reflector and the sixth reflector are arranged in parallel; the second reflector, the third reflector, the fifth reflector and the seventh reflector are arranged in parallel.

[0015] A heat insulation cavity is provided inside the outer shell surface of the frame.

[0016] A heat insulation cavity is provided in the middle of the frame.

[0017] A target for autonomous image marking comprises a target rod, a target base and a target lamp. One end of the target rod is adjustably mounted on the target base. The target lamp is mounted on the target rod and can display light on the front and back surfaces of the target rod.

[0018] As a further improvement of the above technical solution: The target base is provided with a base frame, and a group of mounting and fixing components are respectively provided at both ends of the base frame. The mounting and fixing components include fixing fasteners and a pair of leveling bolts. The target base is provided with a mounting through hole for installing the fixing fasteners, and leveling screw holes are provided on both sides of the mounting through hole for installing a pair of leveling bolts.

[0019] The bottom of the leveling bolt is spherical.

[0020] The base frame is provided with an arc guide groove, the target rod is hinged to the target seat, and the hinge point is located at the center of the arc guide groove. A locking screw is slidably provided on the arc guide groove, and the locking screw cooperates with the target rod to fix the target rod at a specified position.

[0021] There are three target lights, which are evenly spaced along the target pole. The lights displayed by the three target lights on the front and back surfaces are the same in the same direction and different in different directions.

[0022] The surface of the target rod is provided with a low-reflectivity matte coating.

[0023] A target light for a target used for autonomous image identification as described above, comprising a rear scattering cover, a rear hyperbolic reflection cavity, a reflection cone, lamp beads, a front hyperbolic reflection cavity and a front scattering cover, wherein the rear hyperbolic reflection cavity is docked with the front hyperbolic reflection cavity and the reflection cone is clamped and fixed inside, a plurality of lamp beads are provided, and are dispersedly arranged on the front and rear sides of the reflection cone, the front scattering cover is installed at the front end of the front hyperbolic reflection cavity, and the rear scattering cover is installed at the rear end of the rear hyperbolic reflection cavity.

[0024] As a further improvement of the above technical solution: A light-shielding hole is provided in the center of the front scattering cover or the rear scattering cover, and a light-shielding tube is provided at one end of the reflecting cone. The light-shielding tube is sleeved in the light-shielding hole.

[0025] The lamp beads include LED lamp beads of multiple different wavelengths. LED lamp beads of multiple different wavelengths are arranged on both the front and rear sides of the reflective cone. The LED lamp beads of multiple different wavelengths on the same side of the reflective cone are arranged in an alternating manner.

[0026] The lamp beads on the front and rear sides of the reflective cone are arranged alternately.

[0027] The optical centers of the lamp beads on the front and rear sides of the reflection cone are located on the same plane.

[0028] The rear hyperboloid reflection cavity, the front hyperboloid reflection cavity and the reflection cone are coaxially arranged.

[0029] A method for fixing a target for autonomous image marking as described above on an uneven surface comprises the following steps: S1: Pre-installation: Pass the fixing fastener through the installation through hole on the target base and drive it into the uneven surface, and make sure that the fixing fastener does not completely press the target base.

[0030] S2: Leveling: screw the leveling bolts into the leveling screw holes so that the spherical part at the bottom of each leveling bolt contacts and abuts against the uneven surface. Adjust the screwing depth of each leveling bolt so that the posture of the target seat tends to the target installation posture.

[0031] S3: Fixing. When the target base reaches the target installation posture, tighten the fixing fasteners to completely fix the target base and complete the fixed installation of the target on the uneven surface.

[0032] A coding device for target coding comprises a remote host, one or more handheld terminals and one or more target coding modules.

[0033] The handheld terminal is used to communicate with the remote host to synchronize the database, and is also used to communicate with the target coding module to add, delete, check and modify the target coding information.

[0034] The target coding module is integrated into the target and is used to synchronize the real-time coding information of the target to the remote host; it is also used to communicate with the handheld terminal to receive addition, deletion, query and modification instructions from the handheld terminal; and it is also used to send the target's coding and status information to the handheld terminal.

[0035] As a further improvement of the above technical solution: The handheld terminal includes a first radio frequency unit, a first storage unit, a first calculation unit and a WIFI unit.

[0036] The first radio frequency unit includes an RFID tag for communicating with the target encoding module.

[0037] The first storage unit is used to store data of the handheld terminal.

[0038] The first calculation unit is used for calculating coding information of the handheld terminal.

[0039] The WIFI unit is used to communicate with the remote host.

[0040] The target encoding module includes a second radio frequency unit, a second storage unit, a second calculation unit and a NET unit.

[0041] The second radio frequency unit includes an RFID tag, and is used for communicating with the handheld terminal.

[0042] The second storage unit is used to store data of the target encoding module.

[0043] The second calculation unit is used to calculate the coding information of the target coding module.

[0044] The NET unit is used for communicating with the remote host.

[0045] The handheld terminal and the target encoding module communicate with each other through a first radio frequency unit and a second radio frequency unit.

[0046] The network relationship between the remote host and the one or more target encoding modules includes: The remote host can perform unidirectional reading and writing of data on the one or more target encoding modules.

[0047] The network relationship between the remote host and the one or more handheld terminals includes: The remote host can read and write data to the one or more handheld terminals.

[0048] The one or more handheld terminals can read data from the remote host.

[0049] The remote host can publish information to the outside through a mobile network, and the mobile network includes a network provided by a communication service provider.

[0050] The network relationship between the handheld terminal and the target encoding module includes: The handheld terminal can read data from the target encoding module.

[0051] Before the handheld terminal reads data from the target encoding module, the handheld terminal needs to communicate with the remote host to synchronize the database.

[0052] A target coding control method, based on the above-mentioned coding device for target coding, comprises: T1. The handheld terminal communicates with the remote host to synchronize the database, communicates with the target coding module and adds, deletes, checks and modifies the target coding information.

[0053] T2. Communicate with the handheld terminal through the target encoding module to receive add, delete, query, and modify instructions from the handheld terminal; and send the target encoding and status information to the handheld terminal through the target encoding module.

[0054] As a further improvement of the above technical solution: The T1 includes: A1: For targets that need to be initialized or modified, a handheld terminal is used to synchronize the code library with a remote host via a WIFI unit and update the code library in the first storage unit.

[0055] A2: Input the proposed target number through the handheld terminal and encode it through the first calculation unit.

[0056] A3: The first calculation unit reads the data in the first storage unit and compares the input, encoded target number with it. If it is repeated with the existing code, it returns to re-enter; if it is not repeated, the input, encoded target number is written into the first storage unit.

[0057] A4: Trigger the RFID tag of the first radio frequency unit to transmit to the RFID tag of the target second radio frequency unit.

[0058] A5: When the RFID tag of the second radio frequency unit receives the code writing instruction, it triggers the second calculation unit to read and write instructions, and the second storage unit receives and writes the new code.

[0059] A6: The new code written into the second storage unit is synchronized to the remote host through the NET unit, and its code library is updated.

[0060] A7: The second calculation unit outputs the new code of the second storage unit to the corresponding lamp beads of each target light to control the on and off combination of the lamp beads.

[0061] The T2 includes: B1: When the handheld terminal reaches the sensing range of the RFID tag of the second radio frequency unit of the target, the RFID tag of the first radio frequency unit on the handheld terminal communicates and activates with the RFID tag of the second radio frequency unit on the target.

[0062] B2: When the RFID tag of the second radio frequency unit on the target receives a query instruction, the read and write function of the second calculation unit is triggered to read the latest target coding information and working status information in the second storage unit.

[0063] B3: The RFID tag of the second radio frequency unit on the target feeds back the latest target coding information and working status information to the RFID tag of the first radio frequency unit on the handheld terminal, which is then decoded by the first calculation unit and displayed on the handheld terminal.

[0064] The method also includes: reading, writing, collecting and modifying the target's coded information through the remote host, and adjusting the target light on and off combination.

[0065] The reading, writing, collection and modification of the target's coded information by the remote host include: C1: The remote host sends coded information collection instructions to each target.

[0066] C2: Trigger the second computing unit of each target to execute the coding information read and write instructions, and send the coding information stored in the second storage unit to the remote host through the NET unit.

[0067] C3: The remote host verifies the collected target coding information. If there is a duplicate code, the duplicate code is changed, a code change instruction is issued to the corresponding target, and the changed coding information is sent to the corresponding target.

[0068] C4: The second computing unit of the corresponding target receives the code modification instruction sent by the remote host and the modified coding information, and writes the modified coding information into the second storage unit for refreshing.

[0069] C5: The second calculation unit controls the corresponding lamp beads of the target light according to the modified coding information and adjusts the lighting combination.

[0070] Also includes: Status monitoring: When the target light fails, the target is monitored through the encoding device.

[0071] Power failure recovery: When the system is powered off and restored, the target light returns to the default state, and the encoding device is used to restart the target light to restore the target preset state.

[0072] The condition monitoring includes: D1: Optical simulation is used to calculate the most unfavorable lighting distribution for different numbers of damaged lamp beads. The allowable number of damaged lamp beads is determined by comprehensively considering the lighting uniformity and the minimum lighting resolution of the camera.

[0073] D2. The on and off status of each lamp bead is collected at a preset frequency through the second calculation unit, and a corresponding monitoring code is generated; for lamp beads of a certain wavelength in the same target lamp, when the number of damaged lamp beads collected is the number allowed in D1, the lamp bead is still recorded as all lit, otherwise it is recorded as all off; when the lamp beads are marked as all off, the corresponding monitoring code is written into the second storage unit.

[0074] D3. Compare the monitoring code generated in D2 with the original code in the second storage unit. If the comparison results are consistent, it is determined that the wavelength lamp bead is normal and the current state is maintained; if the comparison results are inconsistent, it is determined that the wavelength lamp bead is damaged, and the second calculation unit sends a lighting termination instruction, and all target lights of the target are turned off.

[0075] D4: For the case determined to be damaged in D3, the second storage unit sends the stored monitoring code to the remote host through the NET unit and issues an alarm for maintenance.

[0076] D5: The remote host publishes maintenance information to the off-site terminal via the mobile network.

[0077] The power outage recovery includes: E1: The target coding information is already stored in the second storage unit before power failure.

[0078] E2: The second computing unit detects power-on information and performs target light self-test according to D1 to D5.

[0079] E3: For a target light that has self-tested normally, the second calculation unit outputs the target code in the second storage unit to the corresponding lamp beads of the target light, and restores the on and off combination of the lamp beads.

[0080] A target coding method, based on the above-mentioned target and target light, includes: Assume that there are a target lights on a single target, each target light has b wavelengths of lamp beads, excluding the target light that is completely extinguished, each target light has a total of 2 b -1 combination; and each target has a target lamp, so for a target with a target lamp, each target lamp has b wavelengths, the number of target numbers that can be expressed is: (2 b -1) a , the encoding method is: F1: Number of coding bits for a single target: Each target light has b wavelengths of lamp beads. Each wavelength of lamp beads needs to occupy 1 bit of coding to represent the two states of on or off, where 1 represents on and 0 represents off. Therefore, each target light requires b bits of coding, and a target light of a single target requires a·b bits of coding.

[0081] F2: Single target light coding: For any target light i on the target, the on and off status codes of b wavelengths are sequentially connected into a target light coding to represent the identity of the current target light. Each target light coding has b bits.

[0082] F3: Single target code: Connect the single target light codes from the 1st target light to the bth target light in sequence from left to right to obtain the code of the entire target.

[0083] F4: Convert target number and target code.

[0084] As a further improvement of the above technical solution: The F4 includes: Convert target number to target code: G1: Assume that the user inputs the decimal number of the target to be coded through the handheld terminal or remote host as k 10 , compare with the existing code according to steps A1 to A7. If it is the same as the existing code, return to prompt re-entry, otherwise enter G2.

[0085] G2: Assume that each target has a target light, each target light has b wavelengths, and the decimal number K is calculated by the first calculation unit or the remote host. 10 Convert to 2 b Base code K 2b , where number K 10 The initial value of is: .

[0086] G3: Check K through the first computing unit or remote host 2b Does it contain 0? If it does, then K 10 After adding 1, return to G2 and recalculate; if it does not contain 0, enter G4.

[0087] G4: Calculate K 10 -K0+1, and the decimal number k entered by the user 10 Compare, if K 10 -K0+1=k 10 , then K 10 Convert to binary code K2 and write it into the second storage unit according to the method from A1 to A7; if K 10 -K0+1≠k 10 , then K 10 After adding 1, return to G2 and recalculate.

[0088] The F4 also includes: Convert target code to target number: H1: According to B1 to B3, the handheld terminal or remote host reads the binary code K2 in the second storage unit.

[0089] H2: Convert the binary code K2 to 2 through the first computing unit or the remote host b Base code K 2-2b .

[0090] H3: The first computing unit or the remote host converts the decimal code K 10 Convert to 2 b Base code K 2b , where K 10 The calculation of the initial value K0 is the same as G2.

[0091] H4: K through the first computing unit or remote host 2-2b With K 2b For comparison, if K 2b =K 2-2b , then K 10 -K0+1 output is the decimal number of binary code K2; if K 2b ≠K 2-2b , then K 10After adding 1, return to H3 and recalculate.

[0092] A method for decoding a target light image, based on the above-mentioned equal-intensity parallax-free synchronous image splitting and multi-band imaging device, obtains a target light image set of a target cluster consisting of the above-mentioned target and the above-mentioned target light, and decodes the target light image set, comprising the following steps: R1. Identify the target light spots and target light directions on the images of each wavelength in the target light image set, and remove the target light spots that do not meet the preset requirements.

[0093] R2. Filter the light spots of each wavelength image, merge the light spots from the same target light in each wavelength image to obtain a target light image, and solve the coordinates and area of the bull's eye of the target light image.

[0094] R3. Classify the target lights according to the area of their images, and group the target lights with similar areas as the same target.

[0095] The R1 includes: I1: Binarize the forward and backward target light images respectively, obtain the outer boundary through Canny edge detection, and obtain the spot area; restore the center coordinates of the target light.

[0096] I2: Calculate the variability E of the light spot in the image R , according to the E R Determine the viewing direction.

[0097] I3: For each image, select the light spot with the smallest variation to determine the viewing direction.

[0098] The I1 includes: The forward and backward target light images are binarized separately.

[0099] The identified boundary pixel center is stored as the boundary point (x Bi , j , y Bi , j ), (x Bi , j , y Bi , j ) represents the coordinates of the jth boundary point of the i-th light spot; the total pixel area A inside the boundary is calculated i .

[0100] The center point G coordinate (x Gi , j , y Gi , j ), (x Gi , j ,y Gi , j ) represents the center of the i-th spot. The central dark area is not considered when calculating the backward target light. Gi , j , y Gi , j ): .

[0101] The I2 includes: I21: Calculate the centroid G of the i-th spot i The distance S to the jth boundary point i , j : .

[0102] I22: Calculate the equivalent circle radius R of the i-th spot ei : .

[0103] I23: For the i-th light spot, the equivalent circle radius R ei As a reference, calculate S i , j and the equivalent circle radius R ei Deviation, E Ri That is the variability of the light spot: .

[0104] E Ri The smaller the value, the less the contour change caused by occlusion. When the light spot is a standard circle, the value should be 0.

[0105] The I3 includes: I31: Taking the center of the target light spot as the starting point and the outer boundary of the detection as the limit, calculate the distribution of the gray value k of the image along the x and y directions respectively, and calculate the corresponding gray sample variance S respectively x and S y : .

[0106] .

[0107] I32: Construct the target discriminant: .

[0108] I33: If Δ=0, the target light in the image is determined to be forward-facing; if Δ≠0, the target light in the image is determined to be backward-facing.

[0109] The R1 further comprises: According to the variation E R Combined spot area A i The integrity of the light spot is evaluated and the light spots that meet the following conditions should be eliminated: in is the variation threshold of the unobstructed light spot.

[0110] The decoding method further comprises: Calibrate the variation threshold and determine the initial value of the variation threshold: Q1: Use the light targets arranged on site as calibration objects, and set all light targets to λ through the remote host. k Wavelength, k≤b, where b is the total number of wavelengths that each target light can emit, and all of them are lit.

[0111] Q2: Shooting λ k 1 image at a wavelength, and obtain the area A of each light spot in the image k,i , spot variability E Rk,i ; i≤n, n is the total number of light spots in the image.

[0112] Q3: Keep E Rk,i With A k,i The corresponding relationship between A k,i Arrange into sequence k , n}, and obtain the corresponding spot variation sequence {E Rk,n}.

[0113] Q4: Find the sequence {E Rk,n}'s minimum value term E Rk,m , sequence {A k , n The corresponding item is A k , m .

[0114] Q5: Find the sequence {E Rk,n}'s maximum value term E Rk,p , sequence {A k , n The corresponding item is A k , p .

[0115] Q6: If A in Q5 k , p In the sequence {A k , n} is the smallest, then calculate λ as follows k The area under the wavelength is A k,i The spot variation threshold E R0(k，i) : .

[0116] Q7: If A in Q5 k , p In the sequence {A k , n} is not the smallest, remove the item and re-execute Q5.

[0117] The R2 comprises: J1: Obtain the outer boundary and center coordinates of the target light image of each wavelength, and count the number of light spots in the image of each wavelength respectively. The central dark area of the backward target light is not considered in the analysis.

[0118] J2: For target lights with b wavelengths, the wavelength image with the largest number of light spots is used as the benchmark, and its wavelength is λ t , calculate the distance between the center of each spot image of this wavelength and the center of the spot of other wavelength images: .

[0119] Among them, x k The horizontal coordinate of the center of the kth spot in the wavelength image with the largest number of spots; y k Indicates the vertical coordinate of the center of the kth spot in the wavelength image with the largest number of spots; x λc,n Indicates wavelength is λ c The horizontal coordinate of the center of the nth spot in the wavelength image; λc,n Indicates wavelength is λ c The vertical coordinate of the center of the nth spot in the wavelength image; D(k, λ c , n) represents the kth light spot in the wavelength image with the largest number of light spots and the wavelength λ c The center distance of the nth light spot in the wavelength image.

[0120] J3: At wavelength λ c Among the many points, when a certain point can make the above-calculated D(k, λ c , n) value is the smallest, then the point has the largest number of spots and the wavelength is λ t The k-th point distance is used to achieve overlap; using D(k, λ c , n), the wavelength is established as λ according to the following formula t The correspondence between each target and other wavelength targets: .

[0121] Among them, P(λ t , k) indicates the largest number of light spots with a wavelength of λ t The kth point of P(λ c , q) represents the wavelength λ c The qth point of P(λ t , k) overlap.

[0122] J4: Merge the light spot centers of the same target lights.

[0123] The J4 includes: J41. Equivalently merge the spot centers of different wavelengths of the same target light.

[0124] J42. Merge the spot areas of different wavelengths of the same target light.

[0125] The J41 includes: J411: Set the resolution accuracy: When calculating according to J412 to J418, as the search center gradually approaches the search center point M, the step size will tend to 0; considering the discreteness of digital imaging, Δ≤δ is used as the cutoff condition, where Δ is the step size and δ is the pixel size of the imaging sensor.

[0126] J412: Determine the initial search area: Based on the center point P of the light spot of the same target light with different wavelengths λ1 ~P λm (1≤m≤b) coordinates determine the initial search range, take min(x i )≤x≤max(x i ), min(y i )≤y≤max(y i )(1≤i≤m)area is taken as the initial area.

[0127] J413: Determine the initial search center: J412 determines the geometric center G formed by all points in the area as the initial search center point M, the coordinates are: .

[0128] J414: Determine the initial search step size: half of the minimum value of the height and width directions of the area determined by J412 is used as the initial search step size Δ: .

[0129] J415: Search area: With the optimal point M as the center and Δ as the step size, search points are placed in 8 orthogonal directions around it. For the first calculation, point M is point G; the coordinates of each search point are expressed as: .

[0130] J416: Target calculation: For each search point arranged in J415, calculate the distance between it and the center points of m light spots of the same target light with different wavelengths and sum them: .

[0131] J417: Determine the accuracy based on J411. If the accuracy is met, output point M, which is the optimal point. Otherwise, adjust the search center and step size according to J418.

[0132] J418: For point M that does not meet the conditions of J417, consider the following two situations: Case 1: If M coincides with the center point M in J415, it indicates that the initial search area is too large. In this case, update the step size Δ in J415 to Δ / 2 to further narrow the range; then recalculate according to steps J416 to J417.

[0133] Case 2: If M does not coincide with the center point M in J415, the position of point M should be updated. In this case, point M in step J415 is updated to this point, and then steps J416 to J417 are executed.

[0134] J419: Follow steps J411 to J418 to point P λ1 ~P λm Processing is performed to complete the merging of the centers of light spots of different wavelengths of the same target light.

[0135] The J42 includes: J421: For the light spots of the same target light with different wavelengths, calculate their merged center M (x M , y M ), and then calculate the square of the distance S between the merge center M and the center of each wavelength spot i : .

[0136] J422: Based on the minimum distance deviation D calculated in J416, calculate the distance deviation weight β between the spot center of the same target light with different wavelengths and the merge center M: .

[0137] J423: According to the spot area A of the same target light with different wavelengths i , considering the image of J422 with medium-range deviation weight, calculate the weighted mean of the superposition of the spot areas of each wavelength: .

[0138] J424: Press J421~J423 to point P λ1 ~P λm The corresponding spot areas are processed to complete the merging of the spot areas of different wavelengths of the same target light.

[0139] The R3 comprises: R31: Suppose there are q homologous target lights merged data, and the merge center of each homologous target image is (x Mi , y Mi ), the merged area is A Ci (i≤q), the central dark area of the rearward target light is not considered in the analysis.

[0140] R32: A Ci Rearrange in ascending order.

[0141] R33: From A C1 To start, press A Ci With A Ci+1 Compare and calculate the relative deviation value δ i : .

[0142] R34: If δ i >δ0, then A Ci+1 With A Ci Targets classified into the same category are recorded as category r.

[0143] Otherwise, A Ci+1 The target markers classified into the next category are classified as category r+1.

[0144] Where δ0 is the area resolution.

[0145] R35: According to R31 to R34 above, there are t types of targets in total. Then discard the last type of grouped targets, that is, discard the tth group, and retain the first t-1 groups as the result of target clustering grouping.

[0146] The decoding method further comprises: Calibrate the area resolution and determine the initial value of the area resolution: P1: Use the light targets arranged on the site as calibration objects, and set all light targets to λ through the remote host. k Wavelength, k≤b, where b is the total number of wavelengths that each target light can emit, and all of them are lit.

[0147] P2: Shooting λ k An image under a wavelength, obtain the area A of each light spot in the image k,i , i≤n, n is the total number of light spots in the image.

[0148] P3: A k,i Sort in descending order to get the sequence {A k , n}, n is the total number of spots in the image, and the difference between adjacent items Δ is calculated k,j =A k , j+1 -A k , j ; j≤n, n is the total number of light spots in the image.

[0149] P4: Take δ k , 0= min(Δ k,j ) and δ k ,0>0 is taken as the area resolution of the image at this wavelength.

[0150] P5: Calculate the δ of each spot image under type b wavelength according to P1 to P4 k ,0 value.

[0151] P6: Calculate the variation value of each spot image under the wavelength of type b respectively, and select the spot with the smallest variation value as the calculation object.

[0152] P7: Take the light spot with the smallest wavelength variation value in each category as the calculation object, and calculate the deviation weight β corresponding to each wavelength of the light spot. k .

[0153] P8: The δ calculated in P5 k ,0 value is calculated according to β in P7 k Perform weighted averaging to obtain the area resolution δ0: .

[0154] The decoding method further comprises: The decoding results of the target light image set are integrated into image labels to obtain first data.

[0155] Integrating the decoding results of the target light image set into image tags includes: V1: Record the timestamp and camera number when the image is taken.

[0156] V2: Determine the viewing direction of the image and the available target images for screening, using code 1 to represent forward direction and 0 to represent backward direction.

[0157] V3: Different target lights can be distinguished based on wavelength combination.

[0158] V4: Complete the coordinates of each target light image in the image plane, and use (x i , y i )Record.

[0159] V5: Find all the target lights contained in each target and store the positions of the target lights in the same group as a set.

[0160] V6: Generate a coding sequence using the set in V5 so that each target light image has a unique identifiable number.

[0161] The V6 includes: V61: If the result in V2 is 1, then for each set in V5, sort the x-coordinates of the target lights contained therein in ascending order.

[0162] If V2 is judged to be 0, then for each set in V5, the x-coordinates of the target lights contained therein are sorted in descending order.

[0163] V62: Replace the coordinates corresponding to the target light sequence in V61 with the corresponding target light code in V3 to obtain a new sequence, which is the entire target number.

[0164] The decoding method also includes completing target autonomous identification, the steps are as follows: N1: When the camera is triggered to capture images, each wavelength of the target light image will generate a single-channel image file in the camera and transmit it to the remote host through intra-field communication.

[0165] N2: The remote host calculates and stores the image tag.

[0166] N3: Generate visual overlay images.

[0167] N4: The remote host stores the camera's "Unix timestamp", "camera number" and the "viewing direction", "band number", "target image code" and "target image center" obtained in N2 into a data file DATA.

[0168] N5: Integrate the images of each band, the visual overlay image, and the data file DATA into one TIFF file.

[0169] The N3 includes: N31: Generate a blank image with the same resolution as the images in each band and the color level of each pixel in each channel is 0.

[0170] N32: Traverse each band image according to the following rules and assign a value to each pixel level of the blank image generated by N31: For any pixel (i, j) in the image plane coordinate system of each band image, if the color level value affected by any of the images is 255, the color level of the pixel (i, j) at the corresponding position in the blank image generated by N31 is assigned a value of 255; otherwise, it is assigned a value of 0.

[0171] A spatial multi-target image autonomous identification system, comprising: A target as described above; an imaging device as described above; an encoding device as described above.

[0172] The number of targets arranged is a first preset number, and they are staggered along a preset baseline. The target lights on all targets have the same shape and the same display direction of light; the target lights of adjacent targets on the same baseline are staggered in the shooting direction of the imaging device.

[0173] The number of the imaging devices arranged is a second preset number, and is used to capture the target light image on the target.

[0174] The encoding device is used to set the on and off combination of the target light, encode the target, and decode the target light image collected by the imaging device to complete the autonomous identification.

[0175] A method for autonomous identification of spatial multi-target images, based on the above-mentioned autonomous identification system for spatial multi-target images, comprises: U1. Arrange target clusters: Arrange a first preset number of targets in a staggered manner along a preset baseline. The target lights on all targets have the same shape and display direction of light; the target lights of adjacent targets on the same baseline are staggered in the shooting direction of the imaging device.

[0176] U2. Target coding: The coding device is used to set the on and off combination of the target lights on each target to encode the target.

[0177] U3. Capture and decode target light images: Use the imaging device to capture the target light image on the target to obtain a target light image set; decode the target light image set to obtain decoded data, which includes the target light direction, target light image coordinates, and the number information of the target to which the target light belongs.

[0178] U4. Autonomous identification: associate the decoded data with the target light image set to complete autonomous identification.

[0179] In the U1, there are two or more preset baselines, and the preset baselines are continuous spatial curves on any wall surface in the application scene.

[0180] The present invention has the following beneficial effects: The equal-intensity, parallax-free synchronous image splitting and multi-band imaging device of the present invention allows the light beam to be focused by the lens and then converged on the focal plane behind the lens. First, it is divided into two beams by a first-level spectrometer. The two beams pass through the spectroscopic channel and are split and reflected multiple times before finally arriving in front of multiple imaging components. At this time, the light intensity in front of each imaging component is equal and the image has no parallax; then, the insensitive light beam of the imaging sensor is filtered out by the filter of each imaging component to obtain parallax-free images of different wavelength channels; since the light beams in front of each imaging component are all incident and split through the same lens, the image can be ensured to have no parallax by accurately setting the position of the focusing lens; and by accurately setting the position of the imaging sensor, the light beam can be accurately focused, thereby obtaining a clear, parallax-free image. Since the reflected and transmitted light intensities of the beam splitter are related to the incident angle, it is difficult to ensure that the position and posture of the beam splitter relative to the incident light always remain fixed based on the existing technology. The equal-intensity, parallax-free, synchronous image splitting and multi-band imaging device of the present invention constructs an ingenious optical path. For a light beam of any incident angle emitted from the focal plane, it can ensure that the incident angles of all beam splitters in the entire optical path are consistent, and the incident angles of all reflectors are consistent. An optical path compensation element is provided to ensure that the light intensity is equal when the light beam reaches the imaging component.

[0181] The target for autonomous image identification of the present invention has a target pole for mounting a target lamp which can be adjustably mounted on a target base. When applied to a large-scale cluster multi-target scene, the angle of the target pole can be conveniently adjusted to prevent the target lamps on adjacent targets from blocking each other in the images formed in the camera; and all the target lamps can display lights on the front and back surfaces of the target pole, thereby realizing front and back two-way imaging in conjunction with a camera group in a large-scale cluster multi-target scene.

[0182] The target light of the present invention has a light-shielding hole in the center of the front scattering cover or the rear scattering cover, and a light-shielding tube is provided at one end of the reflective cone. The light-shielding tube is sleeved in the light-shielding hole. The light-shielding tube is used to block part of the reflected light and shape the light spot. The purpose is to achieve different light spots in the front and rear viewing directions of the target to assist the camera in determining the viewing direction. Countersunk holes for installing lamp beads are provided on the front and rear sides of the reflective cone. Each countersunk hole is slightly tilted, and the countersunk holes on the front and rear sides of the reflective cone are horizontally staggered and equal in number. Based on this structure, the optical centers of all the lamp beads on the front and rear sides can be located on the same plane by reasonably setting the hole depth of the countersunk hole. Compared with the existing plane direct-illumination lamp bead layout, this structure achieves large-angle diffusion with light from only a small number of light beads, and ensures uniformity by controlling the aberration through hyperbolic surfaces, thereby avoiding increasing the number of lamp beads to make up for the lack of illumination uniformity, and can effectively reduce equipment heat generation. The target light of the present invention can achieve a large number of target information markings with only a small number of wavelength channels. Each target light contains a combination of three wavelength channels. A combination of n target lights can achieve 3 n With the use of target markers, the amount of information that can be marked and the number of target lights increase exponentially, and the amount of information that can be expressed is huge.

[0183] The target fixing method of the present invention enables the entire target base to contact the uneven surface of the mounting structure through four leveling bolts. The ball heads of the leveling bolts have good adaptability to the uneven surface, making the contact points stable and reliable. After the installation is completed, a stable connection relationship is formed between the entire target base and the mounting structure. When the target base is fixed to a planar structure, there is no need to install the leveling bolts. Therefore, this method can be applied to the surfaces of various structures and has more obvious advantages when installed on structures with uneven surfaces.

[0184] The encoding device for target encoding and the target encoding control method of the present invention can use a remote host and a handheld terminal to encode the target. The remote host can realize large-scale encoding operations, and the handheld terminal can realize fixed-point inspection and deletion and modification of encoding operations, which is easy to operate.

[0185] The target coding method of the present invention is based on the display capability of the target light. It establishes a mapping that can dynamically describe the corresponding relationship between the binary target code and the decimal target number through a conversion function, realizes the accurate conversion function between the binary target code and the decimal target number, facilitates the modification and writing of the code during the use stage, and brings great convenience to the user.

[0186] The target light image decoding method of the present invention determines the viewing direction by determining whether the light spot is circular or annular, while ensuring that the light spot is complete. The set process is: first, an indicator that can describe the difference between the residual image and the complete target image is defined, and a target image with the smallest variation is selected from the image for target type analysis, thereby completing the viewing direction. During the decoding process, there is no need to set up a separate target outside the field for calibration, and calibration can be achieved directly using the target arranged on site. In addition, the variation threshold is adapted to the wavelength and the light spot area, taking into account different bands and different imaging areas. The difference influences are higher, and the judgment accuracy is higher; when the light spots from the same target lamp in each wavelength image are merged, the combination and matching problems of the target lamp images of each wavelength are solved. After the residual image of the target lamp is eliminated, the light spots from the light spots of each wavelength image are screened out as coming from the same target lamp, so that the different wavelength images at the same position can be correctly combined together, and the bull's eye after the multi-band image combination is solved; the so-called merging is not a simple overlap of these points, but an equivalent merging of each point into the optimal point through a constructed algorithm. The characteristics of the optimal point after merging are: the optimal point is consistent with each The sum of known point distances is minimized, ensuring the accuracy of the merged data. When merging the spot areas of the same target light with different wavelengths, the algorithm does not simply add up the areas of each image and then calculate the average. Instead, it takes into account the weight of their position distribution deviations and performs a weighted average of each area. This completes the superposition of multiple band images, further improving the accuracy of the merged data. When classifying target light images on the same target, an algorithm based on the spot area aberration is used to classify target light images with similar areas as the same target, improving the accuracy of the target light image grouping. The above method associates and integrates the target light direction, target light image coordinates, and target number information without the reference of a single image, achieving independent identification of the target image. A TIFF file integrates the images of each band and their jointly processed data. This data processing method can effectively solve the problem of large number of images collected by each camera and the difficulty of storage and classification. Because it contains all information starting from the original image, it has strong traceability and is convenient for later verification and proofreading.

[0187] The spatial multi-target image autonomous identification system and spatial multi-target image autonomous identification method of the present invention include the above-mentioned targets, imaging devices, encoding devices and related control methods, and therefore have corresponding technical effects, which will not be repeated here.

[0188] In addition to the above-described objects, features and advantages, the present invention has other objects, features and advantages. The present invention will be further described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0189] The drawings constituting a part of this application are used to provide a further understanding of the present invention. The illustrative embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0190] Figure 1 This is a schematic diagram of a large-scale cluster of cameras and multiple targets.

[0191] Figure 2 It is a target image identified by a selected camera in a large-scale cluster camera and multi-target scenario.

[0192] Figure 3 It is a target image identified after selecting the dynamic displacement of the camera in a large-scale cluster camera and multi-target scenario.

[0193] Figure 4 This is a schematic diagram of the shooting status of a multi-lens and multi-filter camera.

[0194] Figure 5 This is a schematic diagram of the parallax between multi-band images obtained by a multi-lens and multi-filter camera.

[0195] Figure 6 This is a schematic diagram of the shooting status of a single-lens multi-filter camera.

[0196] Figure 7 This is a schematic diagram of the parallax between multi-band images obtained by a single-lens multi-filter camera.

[0197] Figure 8 This is a schematic diagram of the parallax between multi-band images obtained by an existing optical image splitting device.

[0198] Figure 9 It is a schematic diagram of the first-view stereoscopic structure of the target in the target embodiment of the present invention.

[0199] Figure 10 It is a schematic diagram of the three-dimensional structure of the target from a second perspective in the target embodiment of the present invention.

[0200] Figure 11 It is a schematic diagram of the split three-dimensional structure of the target in the target embodiment of the present invention.

[0201] Figure 12 It is a schematic diagram of the three-dimensional structure of the target rod of the target in the target embodiment of the present invention.

[0202] Figure 13 It is a schematic diagram of the first-view split three-dimensional structure of the target base of the target in the target embodiment of the present invention.

[0203] Figure 14 It is a schematic diagram of the split three-dimensional structure of the target holder of the target embodiment of the present invention from a second perspective.

[0204] Figure 15 1 is a schematic diagram of the three-dimensional structure of a target light in an embodiment of the present invention and a corresponding radial cross-sectional three-dimensional schematic diagram.

[0205] Figure 16 It is a schematic diagram of the disassembled three-dimensional structure of the target light in the embodiment of the present invention.

[0206] Figure 17 It is a schematic diagram of the three-dimensional structure of the reflective cone in the embodiment of the target light of the present invention.

[0207] Figure 18 This is a schematic diagram of the arrangement of lamp beads on the reflective cone in an embodiment of the target lamp of the present invention.

[0208] Figure 19 This is a comparison chart of the lighting effects of the lamp beads in the target lamp embodiment of the present invention and traditional lamp beads.

[0209] Figure 20 1 is a schematic diagram of a target fixation method embodiment of the present invention.

[0210] Figure 21 It is a schematic structural diagram of the target of the present invention being fixed to a mounting structure.

[0211] Figure 22 It is a schematic cross-sectional perspective view of an imaging device embodiment of the present invention.

[0212] Figure 23 It is a schematic cross-sectional structural diagram of an imaging device embodiment of the present invention.

[0213] Figure 24 2 is a schematic diagram of a simulation of an application scenario of an imaging device embodiment of the present invention.

[0214] Figure 25 It is a schematic diagram of a multi-band split image captured by an imaging device according to an embodiment of the present invention.

[0215] Figure 26 It is a control system diagram of an embodiment of the encoding device of the present invention.

[0216] Figure 27 Schematic diagram of the network topology of an encoding device embodiment of the present invention.

[0217] Figure 28 This is a schematic diagram of light distribution when different numbers of lamp beads are damaged.

[0218] Figure 29 Schematic diagram of an embodiment of the encoding method of the present invention.

[0219] Figure 30 It is a schematic diagram of the process of converting numbers into codes in an embodiment of the coding method of the present invention.

[0220] Figure 31 It is a schematic diagram of the process of converting codes into numbers in an embodiment of the coding method of the present invention.

[0221] Figure 32 1 is a schematic diagram of a target light afterimage in an embodiment of a method for decoding a target light image of the present invention.

[0222] Figure 33 Schematic diagram of grayscale distribution of the target light center in an embodiment of the method for decoding the target light image of the present invention.

[0223] Figure 34 It is a schematic diagram of the multi-band image merging principle in an embodiment of the target light image decoding method of the present invention.

[0224] Figure 35 It is a schematic diagram of traversal of the center distance of a multi-band target light image in an embodiment of the target light image decoding method of the present invention.

[0225] Figure 36 It is a schematic diagram of the center merging process of multi-band target light images in an embodiment of the target light image decoding method of the present invention.

[0226] Figure 37 It is a clustering diagram of target light imaging of the same target in the embodiment of the target light image decoding method of the present invention.

[0227] Figure 38 This is a schematic diagram of image label generation in an embodiment of a method for decoding a target light image of the present invention.

[0228] Figure 39 This is a schematic diagram of generating a label image file in an embodiment of a method for decoding a target light image of the present invention.

[0229] The reference numerals in the accompanying drawings represent: 100, target pole; 200, target lamp; 201, backscattering cover; 202, back hyperboloid reflector; 203, reflective cone; 204, lamp beads; 205, front hyperboloid reflector; 206, front scattering cover; 207, light shield; 300, target base; 301, circular arc guide groove; 302, locking screw; 303, chassis; 3031, fixing fastener; 3032, leveling bolt; 3033, mounting hole; 3034, leveling screw hole; 400, lens; 500, rack; 501, spectroscopic channel; 5011, primary spectroscope; 5012, secondary spectroscope; 5013, secondary spectroscope; 5014, residual light absorption cavity; 502, imaging assembly; 5021, focusing lens Mirror; 5022, band filter; 5023, imaging sensor; 503, thermal insulation cavity; 600, remote host; 601, handheld terminal; 6011, first radio frequency unit; 6012, first storage unit; 6013, first computing unit; 6014, WIFI unit; 602, target encoding module; 6021, second radio frequency unit; 6022, second storage unit; 6023, second computing unit; 6024, NET unit; L1, first beam of light; L2, second beam of light; M1, first reflector; M2, second reflector; M3, third reflector; M4, fourth reflector; M5, fifth reflector; M6, sixth reflector; M7, seventh reflector. DETAILED DESCRIPTION

[0230] The embodiments of the present invention are described in detail below with reference to the accompanying drawings. However, the present invention can be implemented in many different ways as defined and covered by the claims.

[0231] Imaging device embodiment: like Figures 22 to 23 As shown, the equal-intensity, parallax-free, synchronous image splitting and multi-band imaging device of this embodiment includes a lens 400 and a frame 500. The frame 500 is mainly used to support and fix optical elements. The lens 400 is installed on the frame 500. The lens 400 is used to focus the light beam of the object to be photographed on the focal plane of the lens 400. A spectroscopic channel 501 and multiple imaging components 502 are provided in the frame 500. The spectroscopic channel 501 serves as a cavity for transmitting the main light path, and each surface of the spectroscopic channel is coated with a light-absorbing coating. The spectroscopic channel 501 transmits the light incident from the lens 400 to each imaging component 502 through multi-level spectroscopic and reflection to form equal-intensity light. The imaging component 502 includes a focusing lens 5021, an imaging sensor 5023 and a band filter 5022 for allowing a light beam of a specific wavelength to pass through. The focusing lens 5021 is arranged in front of the filter 5022 to refocus each beam of light onto the imaging sensor 5023 for imaging. The focusing lenses 5021 are parallel to each other and perpendicular to the optical axis of the lens 400.

[0232] In this embodiment, the imaging assembly 502 comprises four groups. The beam splitting channel 501 comprises a primary beam splitter 5011 and two secondary beam splitters 5012. After the incident light is split into two beams by the primary beam splitter 5011, the two beams are further split into two beams by a secondary beam splitter 5012, resulting in four beams of light that are transmitted to the four groups of imaging assemblies 502 and the beam splitters. In this embodiment, convex lenses that have undergone chromatic aberration correction are used as focusing lenses 5021. The optical performance and dimensions of each focusing lens 5021 are consistent. Chromatic aberration is a type of aberration. After light of different wavelengths is focused by an optical system, the image positions of the respective focusing lenses 5021 are not in the same plane. In this embodiment, the focusing of four wavelengths is required. Due to chromatic aberration, the focusing lenses 5021 cannot focus the light beams of different wavelengths on the same plane. Therefore, chromatic aberration correction is required to reduce this difference. In this embodiment, there are four filters 5022, all perpendicular to the optical axis of the lens 400. The allowable wavelength ranges of the filters 5022 do not overlap. The imaging sensor 5023 is an electrical component used to receive the refocused beams of light and form an image. In this embodiment, there are four imaging sensors 5023, each sensitive only to the permissible wavelengths of the corresponding filters 5022, and used to output images in four different wavelength bands. Each imaging sensor 5023 is perpendicular to the optical axis of the lens 400 and is equipped with a device for precise fine-tuning along its axis to correct chromatic aberration of light of different wavelengths after it is focused. This device images light sources of different wavelengths separately. Therefore, in addition to correcting chromatic aberration of the lens, this embodiment also achieves precise imaging through fine-tuning of the imaging plane.

[0233] In this embodiment, the incident light is dispersed into a first beam of light L1 and a second beam of light L2 by a first-level beam splitter 5011. The first beam of light L1 is reflected by a first reflector M1 and then directed to a second-level beam splitter 5012. After being dispersed into two beams of light by the second-level beam splitter 5012, the two beams are respectively reflected by a second reflector M2 and a third reflector M3 into the corresponding imaging component 502. The second beam of light L2 is directly directed to another second-level beam splitter 5012. After being dispersed into two beams of light by the second-level beam splitter 5012, one beam is reflected by a fourth reflector M4 and a fifth reflector M5 in turn into the corresponding imaging component 502, and the other beam is reflected by a sixth reflector M6 and a seventh reflector M7 in turn into the corresponding imaging component 502.

[0234] In this embodiment, a secondary beam splitter 5013 and a residual light absorption cavity 5014 are provided between the secondary beam splitter 5012 and the imaging component 502. The residual light absorption cavity 5014 is a cavity for absorbing excess transmitted or reflected light generated during the optical path compensation process. The inner surface of the residual light absorption cavity 5014 is coated with a light-absorbing coating to prevent the residual light from interfering with other components when the beam splitters are close to other optical components.

[0235] In this embodiment, the primary beam splitter 5011, secondary beam splitter 5012, and secondary beam splitter 5013 are arranged in parallel. All beam splitters are used to split an incident beam into a reflected beam and a transmitted beam. The optical performance and thickness of the beam splitters in this embodiment are completely consistent. The first reflector M1, fourth reflector M4, and sixth reflector M6 are arranged in parallel and parallel to the beam splitters. The second reflector M2, third reflector M3, fifth reflector M5, and seventh reflector M7 are arranged in parallel and perpendicular to the beam splitter. The optical performance of all reflectors is completely consistent.

[0236] In this embodiment, multiple insulation cavities 503 are provided inside and in the middle of the outer shell surface of the rack 500. The multiple insulation cavities 503 are separated by multiple stiffening ribs. The stiffening ribs, as subordinate components of the rack 500, can increase the rigidity of the rack 500 and ensure that the rack 500 does not deform. The insulation cavities 503 can achieve a thermal insulation effect, reduce deformation of the rack 500 caused by temperature changes, prevent the position of the optical components from changing, and ensure the stability of the optical path.

[0237] The equal-intensity, parallax-free synchronous image splitting and multi-band imaging device of the present invention focuses the light beam through the lens 400 and converges on the focal plane behind the lens. First, it is split into two beams by a first-level spectrometer 5011. The two beams pass through the spectrometer channel 501 and undergo multiple splitting and reflections before finally arriving in front of multiple imaging components 502. At this time, the light intensity in front of each imaging component 502 is equal and the image has no parallax. Subsequently, the filter 5022 of each imaging component 502 is used to filter out the insensitive light beam of the imaging sensor 5023, and obtain parallax-free images of different wavelength channels. Since the light beams in front of each imaging component 502 are all incident and split through the same lens 400, the image can be ensured to have no parallax by accurately setting the position of the focusing lens 5021. By accurately setting the position of the imaging sensor 5023, the light beam can be accurately focused, thereby obtaining a clear image with no parallax (such as Figure 24 、 Figure 25 As shown in the figure, the reflected and transmitted light intensities of the beam splitter are related to the incident angle. It is difficult to ensure that the beam splitter's position and posture relative to the incident light remain fixed based on existing technologies. The present invention's uniform light intensity, parallax-free, synchronous image splitting and multi-band imaging device employs a clever optical path. This ensures that the incident angles of all beam splitters and all reflectors in the optical path are consistent for light beams emitted from the focal plane at any incident angle. Furthermore, an optical path compensation element is provided to ensure that the light beams have equal intensity when they reach imaging assembly 502.

[0238] Target Examples: like Figures 9 to 12As shown, the target for autonomous image identification of this embodiment includes a target rod 100, a target base 300 and a target light 200. One end of the target rod 100 can be adjustably mounted on the target base 300, and the angle can be adjusted relative to the mounting surface of the target base 300. The target light 200 is installed in a hole reserved on the target rod 100, and can display light on the front and back surfaces of the target rod 100. The target rod 100 is formed in one step using a lightweight aluminum alloy, and a low-reflectivity matte coating is sprayed on the surface to prevent the light source emitted by the target light 200 from being reflected by it and captured by the camera to affect the image quality. The target for autonomous image identification of the present invention has a target pole 100 for mounting a target light 200 which can be adjustably mounted on a target base 300. When used in a large-scale cluster multi-target scene, the angle of the target pole 100 can be conveniently adjusted to prevent the target lights 200 on adjacent targets from blocking each other's images in the camera; and all the target lights 200 can display lights on the front and back surfaces of the target pole 100, thereby realizing front and back two-way imaging in conjunction with a camera group in a large-scale cluster multi-target scene.

[0239] like Figure 13 、 Figure 14 As shown, in this embodiment, the target base 300 is provided with a base frame 303, which provides support for the rotation of the target rod and the connection with the structure, and is formed in one piece using a lightweight aluminum alloy; a group of mounting and fixing components are respectively provided at both ends of the base frame 303, and the mounting and fixing components include a fixing fastener 3031 and a pair of leveling bolts 3032. The target base 300 is provided with a mounting through hole 3033 for installing the fixing fastener 3031, and leveling screw holes 3034 are provided on both sides of the mounting through hole 3033 for installing a pair of leveling bolts 3032. The bottom of the leveling bolt 3032 is spherical, and the fixing fastener 3031 is an expansion bolt. During installation, a pair of leveling bolts 3032 are screwed into the leveling screw holes 3034, and the spherical part at the bottom of the leveling bolts 3032 abuts against the surface of the mounting structure. By adjusting the screwing depth of the pair of leveling bolts 3032 and the leveling screw holes 3034, the gap and distance between the base frame 303 and the mounting structure can be controlled. After the base frame 303 is basically in the required installation state, the target base 300 is fixed to the surface of the mounting structure by installing the fixing fasteners 3031.

[0240] like Figure 13 、 Figure 14 As shown, in this embodiment, an arc guide groove 301 is provided on the base frame 303, the target rod 100 is hinged to the target base 300, and the hinge point is located at the center of the arc guide groove 301, and a locking screw 302 is slidably provided on the arc guide groove 301. The locking screw 302 cooperates with the target rod 100 to fix the target rod 100 in a specified position. The locking screw 302 includes a butterfly nut, and whether the target rod 100 can rotate is controlled by tightening or loosening the butterfly nut.

[0241] In this embodiment, there are three target lights 200, which are evenly spaced along the target pole 100. The three target lights 200 are all light sources that can provide multi-bands, and the lights displayed by the three target lights 200 on the front and back surfaces are the same in the same direction and different in different directions.

[0242] Target light embodiment: like Figures 15 to 18 As shown, the target light of this embodiment includes a rear scattering cover 201, a rear hyperbolic reflection cavity 202, a reflection cone 203, lamp beads 204, a front hyperbolic reflection cavity 205 and a front scattering cover 206. The rear hyperbolic reflection cavity 202 is connected to the front hyperbolic reflection cavity 205 and clamps the reflection cone 203 inside. There are multiple lamp beads 204, which are dispersed on the front and back sides of the reflection cone 203. The front scattering cover 206 is installed at the front end of the front hyperbolic reflection cavity 205, and the rear scattering cover 201 is installed at the rear end of the rear hyperbolic reflection cavity 202. The rear scattering cover 201 scatters the light reflected backward and makes it uniform. The front scattering cover 206 scatters the light reflected forward and makes it uniform. The rear hyperboloid reflective cavity 202 and the front hyperboloid reflective cavity 205 are both made of light aluminum alloy and processed in one go to facilitate the heat dissipation of the light source. The rear hyperboloid reflective cavity 202 shapes the light path according to a certain rule and then reflects it to the rear of the target. The front hyperboloid reflective cavity 205 shapes the light path according to a certain rule and then reflects it to the front of the target. The reflective cone 203 The emitted light is deflected in the forward and backward directions in the reflection cavity. The inner side of the rear scattering cover 201 and the front scattering cover 206 is spherical, and the outer side is flat. Both the inner and outer sides are treated with a frosted process. A slot for installing the rear scattering cover 201 is provided on the hyperbolic reflection cavity 202, and the slot wall is sprayed with a low-reflectivity matte coating; the inner surfaces of the rear hyperbolic reflection cavity 202 and the front hyperbolic reflection cavity 205 are both curved surfaces formed by rotating a hyperbola around the central axis of the target lamp, and their surfaces are metal-coated mirrors.

[0243] In any embodiment, a light-shielding hole is opened in the center of the front scattering cover 206 or the rear scattering cover 201, and a light-shielding tube 207 is provided at one end of the reflection cone 203, and the light-shielding tube 207 is sleeved in the light-shielding hole. In this embodiment, a light-shielding hole is opened on the rear scattering cover 201. The light-shielding tube 207 on the reflection cone 203 is an inverted cone structure, located at the top of the rear side of the reflection cone 203, and cooperates with the light-shielding hole on the rear scattering cover 201. The light-shielding tube 207 is used to block part of the reflected light and shape the light spot. The purpose is to achieve different light spots in the front and rear viewing directions of the target, so as to assist the camera in determining the viewing direction.

[0244] In any embodiment, the lamp beads 204 include LED lamp beads of multiple different wavelengths, and LED lamp beads of multiple different wavelengths are provided on the front and back sides of the reflection cone 203. The LED lamp beads of multiple different wavelengths on the same side of the reflection cone 203 are arranged in an staggered manner. In this embodiment, the lamp beads 204 are LED lamp beads of three different wavelengths: red, green, and blue. The wavelengths of the selected LED lamp beads have good penetrability and meet the sensitivity difference requirements of the sensor.

[0245] In this embodiment, the lamp beads 204 on the front and rear sides of the reflective cone 203 are arranged in a staggered manner, with the optical centers of the lamp beads 204 on the front and rear sides of the reflective cone 203 located on the same plane. The rear hyperboloid reflective cavity 202, the front hyperboloid reflective cavity 205, and the reflective cone 203 are arranged coaxially. Countersunk holes for mounting the lamp beads 204 are provided on both the front and rear sides of the reflective cone 203. Each countersunk hole is slightly tilted, and the countersunk holes on the front and rear sides of the reflective cone 203 are arranged in a horizontally staggered manner, with an equal number. Based on this structure, by properly setting the hole depth of the countersunk holes, the optical centers of all the lamp beads 204 on the front and rear sides can be located on the same plane. Compared to existing planar direct-beam lamp bead layouts, this structure achieves wide-angle diffusion with only a small number of light beads, and ensures uniformity by controlling aberrations through the hyperboloid, thus avoiding the need to increase the number of lamp beads to compensate for insufficient illumination uniformity and effectively reducing device heat generation. A tray is provided on the outer periphery of the reflection cone 203 , and the tray is used to cooperate with the rear hyperboloid reflection cavity 202 and the front hyperboloid reflection cavity 205 to limit and fix the entire reflection cone 203 .

[0246] Based on the target lamp structure of the present invention, through the cooperation of the rear scattering cover 201 and the rear hyperbolic reflective cavity 202, and the cooperation of the front scattering cover 206 and the front hyperbolic reflective cavity 205, the point light source formed by each lamp bead 204 can be made equivalent to a single point light source, so that the light emitted by each lamp bead 204 has the same optical path after being reflected by the inner surface of the reflective cavity and reaching the scattering cover, so that it has good spatial homology and uniformity. Figure 19 As shown in the figure, through optical simulation software simulation, the lighting effect is more uniform than that of traditional lamp beads.

[0247] Based on the three different wavelengths of red, green and blue LED lamp beads in this embodiment, by controlling the on and off combinations of the different wavelength lamp beads 204, a single target light can form seven different combination output results, as shown in the following table: The combined output results of the front and rear sides of the target light have the same color but different shapes. Since the light shielding hole on the backscatter cover 201 cooperates with the light shielding tube 207 on the reflective cone 203 to block part of the reflected light, the light spot formed on the rear side of the target light is an annular light spot, while the front side of the target light is not blocked, so it is a circular light spot, which can realize the rapid determination of the camera's viewing direction; the target light of this embodiment can realize a large amount of target information marking through only a small number of wavelength channels. Each target light contains 7 wavelength channel combinations. n target light combinations can realize 7 n With the use of target markers, the amount of information that can be marked and the number of target lights increase exponentially, and the amount of information that can be expressed is huge.

[0248] In practical applications, after all targets are pre-encoded by the encoding device of the present invention, the on and off states of the target lights on each target are uniformly set and allocated. The specific encoding device, encoding method, etc. are described in the following embodiments.

[0249] Target fixation method example: This embodiment is a method for fixing a target for image autonomous marking on an uneven surface based on the above target embodiment. The flow chart of this method is as follows: Figure 20 As shown, the installation structure is as follows Figure 21 As shown, the specific steps include: S1: Pre-installation: Pass the fixing fastener 3031 through the mounting hole 3033 on the target base 300 and drive it into the uneven surface. The fixing fastener 3031 does not completely press the target base 300. Since the bottom surface of the target base 300 is flat, it cannot be tightly attached to the surface of the mounting structure. There is an unstable contact point between the bottom surface of the target base 300 and the surface of the mounting structure. At this time, the target base 300 is still prone to displacement under external disturbances. From then on, the next step of installation is entered.

[0250] S2: Leveling: screw the leveling bolts 3032 into the leveling screw holes 3034 so that the spherical portion at the bottom of each leveling bolt 3032 contacts and abuts against the uneven surface. Adjust the screwing depth of each leveling bolt 3032 so that the posture of the target base 300 approaches the target installation posture.

[0251] S3: Fixing. When the target base 300 reaches the target installation posture, tighten the fixing fastener 3031 to completely fix the target base 300, thereby completing the fixed installation of the target on the uneven surface.

[0252] By adopting the method of this embodiment, the entire target base 300 is brought into contact with the uneven surface of the mounting structure through the four leveling bolts 3032. The ball heads of the leveling bolts 3032 have good adaptability to the uneven surface, making the contact points stable and reliable. After the installation is completed, a stable connection is formed between the entire target base 300 and the mounting structure. When the target base 300 is fixed to a planar structure, there is no need to install the leveling bolts 3032. Therefore, this method is applicable to the surfaces of various structures and has a more obvious advantage when installed on structures with uneven surfaces.

[0253] Encoding device embodiment: like Figure 26 、 Figure 27 As shown, the encoding device for target encoding of this embodiment includes a remote host 600, one or more handheld terminals 601 and one or more target encoding modules 602; the handheld terminal 601 is used to communicate with the remote host 600 to synchronize the database, and is also used to communicate with the target encoding module 602 and add, delete, query and modify the target encoding information; the target encoding module 602 is integrated in the target, and is used to synchronize the real-time encoding information of the target to the remote host 600, communicate with the handheld terminal 601 to receive the addition, deletion, query and modification instructions from the handheld terminal 601, and send the target's encoding and status information to the handheld terminal 601.

[0254] In this embodiment, the handheld terminal 601 includes a first radio frequency unit 6011, a first storage unit 6012, a first computing unit 6013, and a Wi-Fi unit 6014. The first radio frequency unit 6011 includes an RFID tag for communicating with the target encoding module 602. The first storage unit 6012 is used to store data on the handheld terminal 601. The first computing unit 6013 is used to calculate encoding information for the handheld terminal 601. The Wi-Fi unit 6014 is used to communicate with the remote host 600. The main function of the handheld terminal 601 is to initialize and modify the encoding during on-site installation and commissioning of the target. Synchronization with the remote host 600 database via the Wi-Fi unit 6014 ensures the accuracy and real-time performance of the handheld terminal 601 database. Communication with the target encoding module 602 via the first radio frequency unit 6011 enables on-site addition, deletion, query, and modification of target encoding information.

[0255] In this embodiment, the target encoding module 602 is independently integrated into the target and includes a second radio frequency unit 6021, a second storage unit 6022, a second calculation unit 6023, and a network unit 6024. The second radio frequency unit 6021 includes an RFID tag for communicating with the handheld terminal 601. The second storage unit 6022 is used to store data from the target encoding module 602. The second calculation unit 6023 is used to calculate the encoding information of the target encoding module 602. The network unit 6024 is used to communicate with the remote host 600. The handheld terminal 601 and the target encoding module 602 communicate via the first radio frequency unit 6011 and the second radio frequency unit 6021, and this communication is achieved via the local area network. The NET unit 6024 synchronizes the target's real-time encoding information to the remote host 600. The second radio frequency unit 6021 communicates with the first radio frequency unit 6011 of the handheld terminal 601, accepting code deletion and modification instructions from the handheld terminal 601 and uploading the target's code and status information to the handheld terminal 601.

[0256] In this embodiment, the network relationship between the remote host 600 and one or more target encoding modules 602 includes: the remote host 600 can perform unidirectional data reading and writing on one or more target encoding modules 602. The network relationship between the remote host 600 and one or more handheld terminals 601 includes: the remote host 600 can read and write data on one or more handheld terminals 601; each handheld terminal 601 can read data from the remote host 600. The remote host 600 can publish information to the outside through a mobile network, and the mobile network includes the network provided by the communication service provider. The network relationship between the handheld terminal 601 and the target encoding module 602 includes: the handheld terminal 601 can read data from the target encoding module 602. Before the handheld terminal 601 reads data from the target encoding module 602, the handheld terminal 601 needs to communicate with the remote host 600 to synchronize the database.

[0257] Target coding control method embodiment: The target coding control method of this embodiment is based on the aforementioned coding device and target light embodiments, and the method includes: T1. The handheld terminal 601 communicates with the remote host 600 to synchronize the database, and communicates with the target coding module 602 to add, delete, check and modify the target coding information.

[0258] T2. Communicate with the handheld terminal 601 through the target coding module 602 to receive the add, delete, query and modify instructions from the handheld terminal 601; send the target code and status information to the handheld terminal 601 through the target coding module 602.

[0259] In this embodiment, T1 and T2 are mainly used for querying, initializing, and modifying the target code on site, wherein T1 includes: A1: For a target that needs to initialize or modify the code, the handheld terminal 601 is used to synchronize the code library with the remote host 600 via the WIFI unit 6014 and update the code library in the first storage unit 6012.

[0260] A2: Input the proposed target number through the handheld terminal 601 and encode it through the first calculation unit 6013.

[0261] A3: The first calculation unit 6013 reads the data in the first storage unit 6012 and compares the input, encoded target number with it. If it is repeated with the existing code, it returns to re-enter; if it is not repeated, the input, encoded target number is written into the first storage unit 6012.

[0262] A4: Trigger the RFID tag of the first radio frequency unit 6011 to transmit to the RFID tag of the target second radio frequency unit 6021.

[0263] A5: When the RFID tag of the second radio frequency unit 6021 receives the code writing instruction, it triggers the second calculation unit 6023 to read and write instructions, and the second storage unit 6022 receives and writes the new code.

[0264] A6: The new code written into the second storage unit 6022 is synchronized to the remote host 600 through the NET unit 6024, and its code library is updated.

[0265] A7: The second calculation unit 6023 outputs the new code of the second storage unit 6022 to the corresponding lamp bead 204 of each target light to control the on and off combination of the lamp bead 204.

[0266] In this embodiment, step T2 includes: B1: When the handheld terminal 601 reaches the sensing range of the RFID tag of the second RF unit 6021 of the target, the RFID tag of the first RF unit 6011 on the handheld terminal 601 communicates and activates with the RFID tag of the second RF unit 6021 on the target.

[0267] B2: When the RFID tag of the second radio frequency unit 6021 on the target receives a query instruction, the read and write functions of the second calculation unit 6023 are triggered to read the latest target coding information and working status information in the second storage unit 6022.

[0268] B3: The RFID tag of the second radio frequency unit 6021 on the target feeds back the latest target coding information and working status information to the RFID tag of the first radio frequency unit 6011 on the handheld terminal 601, which is then decoded by the first calculation unit 6013 and displayed on the handheld terminal 601.

[0269] In this embodiment, the target's coding information can also be read, written, collected, and modified by the remote host 600, and the target light on / off combination can be adjusted. This part of the method is mainly used for remote, large-scale collection and modification of target coding, specifically including: C1: The remote host 600 sends a coded information collection instruction to each target.

[0270] C2: Trigger the second computing unit 6023 of each target to execute the coding information read and write instruction, and send the coding information stored in the second storage unit 6022 to the remote host 600 through the NET unit 6024.

[0271] C3: The remote host 600 checks the collected target coding information. If there is a duplicate code, the duplicate code is changed, a code change instruction is issued to the corresponding target, and the changed coding information is sent to the corresponding target.

[0272] C4: The second computing unit 6023 of the corresponding target receives the code change instruction sent by the remote host 600 and the code information after the code change, and writes the code information after the code change into the second storage unit 6022 for refreshing.

[0273] C5: The second calculation unit 6023 controls the corresponding lamp beads 204 of the target light according to the modified coding information and adjusts the lighting combination thereof.

[0274] The target's information is expressed by the combination of the on and off lights 204. If the lights 204 malfunction and the lights are inconsistent with the code, such as if a certain wavelength of the lights 204 is damaged, the wavelength combination of the image will be inconsistent with the actual code, causing the system to err in encoding and decoding the image. When the system loses power, the target light needs to be re-lit after the power is restored, and it needs to be restored to the preset state. This embodiment also includes two control modes: status monitoring and power failure recovery. Status monitoring: When the target light fails, the target state is monitored through the encoding device. Power failure recovery: When the system loses power and the target light returns to its default state, the encoding device restarts the target light and restores the target to its preset state.

[0275] In this embodiment, the specific steps of status monitoring include: D1: Using optical simulation, the most unfavorable illumination distribution for different numbers of damaged lamp beads 204 is calculated. The allowable number of damaged lamp beads 204 is determined by comprehensively considering illumination uniformity and the minimum illumination resolution of the camera. Figure 28 This diagram illustrates the illumination distribution for different numbers of damaged lamp beads 204 when there are four single-wavelength lamp beads 204. As shown in Figure (c), when two lamp beads 204 are damaged (the figure represents the most unfavorable operating condition, where two adjacent lamp beads are damaged and the illumination is severely eccentric), a significant low-illuminance region appears in the illumination distribution. Therefore, when there are four single-wavelength lamp beads 204, one damaged lamp bead 204 is allowed. This method takes into account engineering practicality and economic efficiency, allowing a certain number of damaged lamp beads 204 of the same wavelength in each target light. This allowable number of damaged lamp beads 204 is determined using the aforementioned optical simulation method.

[0276] D2. The on and off status of each lamp bead 204 is collected at a preset frequency through the second calculation unit 6023, and a corresponding monitoring code is generated; for the lamp bead 204 of a certain wavelength in the same target lamp, when the number of damaged lamp beads 204 collected is the number allowed in D1, the lamp bead 204 is still recorded as all lit, otherwise it is recorded as all extinguished; when the lamp bead 204 is marked as all extinguished, the corresponding monitoring code is written into the second storage unit 6022; taking D1 as an example, when four lamp beads 204 of a certain wavelength are in a lighting combination, one is detected to be damaged. At this time, the system still considers that all four are lit, and its corresponding code is recorded as 1, and the monitoring code is not written into the second storage unit 6022; if two lamp beads 204 are detected to be damaged, the system considers that all four lamp beads 204 are extinguished, and its corresponding code is recorded as 0, and the monitoring code is written into the second storage unit 6022.

[0277] D3. Compare the monitoring code generated in D2 with the original code in the second storage unit 6022. If the comparison results are consistent, it is determined that the wavelength lamp bead 204 is normal and the current state is maintained; if the comparison results are inconsistent, it is determined that the wavelength lamp bead 204 is damaged, and the second calculation unit 6023 sends a lighting termination instruction, and all target lights of the target are turned off.

[0278] D4: For the case determined to be damaged in D3, the second storage unit 6022 sends the stored monitoring code to the remote host 600 through the NET unit 6024 and issues an alarm for maintenance.

[0279] D5: The remote host 600 publishes the maintenance information to the off-site terminal via the mobile network.

[0280] In this embodiment, it is important to note that regarding power outage recovery, as can be seen from the aforementioned control logic, when encoding the target, the handheld terminal 601 and the remote host 600 may simultaneously read and write target data (the remote host 600 has a much higher read and write frequency than the handheld terminal 601). To ensure the accuracy and reliability of the code stored in the target, in this embodiment, the target's RFID tag is only used for communication with the handheld terminal 601 and temporary storage during encoding. The target's RFID tag is not used as a permanent storage medium. The data processed by the target's second computing unit 6023 is stored in a separate second storage unit 6022. Therefore, if power is lost within the field, upon power restoration, the target's preset state is restored by reading the second storage unit 6022, rather than the target's RFID tag.

[0281] The specific steps of power failure recovery in this embodiment include: E1: Before power failure, the second storage unit 6022 has stored target coding information.

[0282] E2: The second calculation unit 6023 detects the power-on information and performs a target light self-check according to D1 to D5.

[0283] E3: For a target light that has self-tested normally, the second calculation unit 6023 outputs the target code in the second storage unit 6022 to the corresponding lamp bead 204 of the target light, and restores the on and off combination of the lamp bead 204.

[0284] Target encoding method embodiment: A target coding method is based on the above-mentioned target and target lamp embodiment. Assume that there are a target lamps on a single target, each target lamp has lamp beads 204 of b wavelengths, and excluding the target whose target lamp is completely extinguished, each target lamp has a total of 2 b -1 combination; and each target has a target lamp, so for a target with a target lamp, each target lamp has b wavelengths, the number of target numbers that can be expressed is: (2 b -1) a In this embodiment, a=3, b=3, and each target light has 2 3 -1=7 combinations, the number of target numbers that can be expressed is: (2 3 -1) 3 =343 kinds.

[0285] like Figures 29 to 31 As shown, the target encoding method includes the following steps: F1: Number of coding bits for a single target: Each target light has b wavelength lamp beads 204. Each wavelength lamp bead 204 requires 1 bit of coding to represent two states of on or off, where 1 represents on and 0 represents off. Therefore, each target light requires b bits of coding, and a target light of a single target requires a·b bits of coding. In this embodiment, each target light is coded with 3 bits, and the 3 target lights of a single target are coded with 3·3=9 bits.

[0286] F2: Single target light coding: For any target light i on the target, the on / off status codes of b wavelengths are sequentially connected into a target light coding to represent the identity of the current target light. Each target light coding has b bits; Figure 29 The target light 2 has three wavelengths λ1, λ2, λ b The on and off states are 1, 1, 1 respectively. When connected into a single target light, the code is 111.

[0287] F3: Single target code: Connect the codes of the single target lights from the first target light to the bth target light in order from left to right to obtain the code of the entire target. In this embodiment, this order is globally unified. Even if the code is reversed during the backward coding, the coded image is read in this order during decoding and connected in order to obtain the code of the entire target. Figure 29 In the forward target, the codes of the three target lights are 010, 111, and 100 respectively. After connecting them in sequence, the code of the target is 010111100; since the wavelength combination of the rearward target light is consistent with that of the forward target light, the position of the target light in the image will be reversed, and the target code displayed is 100111010.

[0288] F4: Convert target number and target code.

[0289] Since the target code is binary, it is suitable for computer processing, but it is not convenient for modification and writing in the stage code. Therefore, we need to design a transcoding method that meets the following conditions: ⅰ The numbers entered by the user are consecutive decimal numbers; ⅱ The number entered by the user corresponds one-to-one with the code in F3. However, it should be noted that not all decimal numbers can be mapped to the code in F3. This is because when a decimal number is converted to binary, there may be consecutive zeros. For example, 400 is converted to binary as 100 000 000. According to the F3 coding rules, only the first target light is lit, and the other two are off. This is not allowed. Therefore, this situation should be excluded when establishing the mapping relationship.

[0290] iii. The correspondence mentioned in ii should not be a search of a pre-built table, but rather a mapping that dynamically describes the correspondence, such as a conversion function. Because pre-built tables rely on an exhaustive approach, they are less flexible, require a large amount of storage, and are difficult to maintain. Furthermore, they fail to fundamentally reflect the correspondence and fail to resolve the decimal code discontinuity issue mentioned in ii. This will cause the user terminal to avoid some numbers, causing significant inconvenience.

[0291] To address the above problem, step F4 of this embodiment includes two parts: one is converting the target number into the target code, and the other is converting the target code into the target number.

[0292] In this embodiment, converting the target number into the target code includes: G1: Assume that the user inputs the decimal number of the target to be encoded as k through the handheld terminal 601 or the remote host 600 10 , compare with the existing code according to steps A1 to A7. If it is the same as the existing code, return to prompt re-entry, otherwise enter G2.

[0293] G2: Assume that each target has a target light, and each target light has b wavelengths. The first calculation unit 6013 or the remote host 600 calculates the decimal number K 10 Convert to 2 b Base code K 2b In this embodiment, each target lamp has three wavelengths, so K 10 Convert to 2 3 = octal code K8, where number K 10 The initial values are: .

[0294] G3: Check K by the first computing unit 6013 or the remote host 600 2b Does it contain 0? If it does, then K 10 After adding 1, return to G2 and recalculate; if it does not contain 0, enter G4.

[0295] G4: Calculate K 10 -K0+1, and the decimal number k entered by the user 10 Compare, if K 10 -K0+1=k 10 , then K 10 Convert to binary code K2 and write into the second storage unit 6022 according to the method of steps A1 to A7; if K 10 -K0+1≠k 10 , then K 10 After adding 1, return to G2 and recalculate.

[0296] The essence of the above method is in 2 b Under binary encoding, find the kth code K that does not contain 0 2b , and k is used as a binary coding mapping, which not only meets the target light coding rules but also ensures the continuity of the numbering.

[0297] In this embodiment, converting the target code into the target number includes: H1: According to B1 to B3, the handheld terminal 601 or the remote host 600 reads the binary code K2 in the second storage unit 6022.

[0298] H2: The first computing unit 6013 or the remote host 600 converts the binary code K2 into 2 b Base code K 2-2b .

[0299] H3: The first computing unit 6013 or the remote host 600 converts the decimal code K 10 Convert to 2 b Base code K 2b , where K 10 The calculation of the initial value K0 is the same as G2.

[0300] H4: K is processed by the first computing unit 6013 or the remote host 600 2-2b With K 2b For comparison, if K 2b =K 2-2b , then K 10 -K0+1 output is the decimal number of binary code K2; if K 2b ≠K 2-2b , then K 10 After adding 1, return to H3 and recalculate.

[0301] The essence of the above method is in 2 b Under the binary encoding, it is known that its code is K 2b , find out which code does not contain 0. Its advantage is that the conversion between numbers and codes can be completed using only the system's RAM, without occupying the space of the remote host 600, the handheld terminal 601, and the target's second storage unit 6022.

[0302] To facilitate the understanding of the above method, the following table provides a section number and code of a target containing three target lights and three wavelengths for reference.

[0303] As can be seen from the table above, converting decimal to 2 b The base is actually to simplify the judgment of multiple sub-codes in binary into the judgment of one code. It can also calculate the number of unencodable codes N as: .

[0304] The aforementioned encoding method uses the forward-facing target light as the reference, and input and decoding are performed from left to right. For the rearward-facing target light, since its wavelength combination is consistent with the corresponding forward-facing target light, the influence of forward and backward directions is not considered during input and decoding. The reverse encoding process of the rearward-facing target light is completed during the image decoding stage.

[0305] Target light image decoding method embodiment: A method for decoding a target light image is provided. Based on the above-mentioned embodiment of the iso-intensity, parallax-free synchronous image splitting and multi-band imaging device, a target light image set of a target cluster consisting of the above-mentioned target and target light embodiment is obtained, and the target light image set is decoded. The method specifically comprises the following steps: R1. Identify the target light spots and target light directions on the images of each wavelength in the target light image set, and eliminate the target light spots that do not meet the preset requirements.

[0306] R2. Filter the light spots of each wavelength image, merge the light spots from the same target light in each wavelength image to obtain the target light image, and solve the target center coordinates and area of the target light image.

[0307] R3. Classify the target lights according to the area of their images, and group the target lights with similar areas as the same target.

[0308] In this embodiment, when the imaging device captures a target image within the visual field, its viewing direction is not fixed. If the viewing direction is reversed, the captured target image will also be reversed. If this is not distinguished, recognition errors will occur according to the aforementioned encoding rules. To this end, the aforementioned target light is designed with inconsistent forward and backward directions (a circular light spot in the forward direction and an annular light spot in the backward direction) to facilitate differentiation of the imaging device's viewing direction from the image.

[0309] Furthermore, during the execution of step R1, the target will inevitably be blocked from the front and back, resulting in incomplete imaging of the target light spot. In this case, the light spot does not appear elliptical, but rather has clipped edges, which we call an afterimage. Based on the spatial position of the target, the target image is mainly divided into the following three situations: the first is corner occlusion: this is mainly manifested by the target image light spot being blocked by the corner of another target, and the overall structure is a concave polygon; the second is majority occlusion: this is mainly manifested by the absence of most of the target image light spot, revealing only a crescent-shaped light spot structure; the third is minor occlusion: this is mainly manifested by the absence of a small portion of the target image light spot, revealing only a convex moon-shaped light spot structure.

[0310] Since the forward and backward light spots have different shapes, the forward one is circular and the backward one is annular, and the aforementioned occlusion affects the judgment of the light spot type (circular or annular), regardless of whether it is a forward or backward light spot, since there is no background reference, the above-mentioned afterimages will interfere with the subsequent analysis of the light spot. Therefore, they must be removed from the identified light spot and not participate in the analysis. Therefore, it is necessary to ensure that the light spot is complete before determining whether it is circular or annular to determine the viewing direction. The set process is: first define an indicator (variability) that can describe the difference between the afterimage and the complete target image, and select a target image with the smallest variation from the image (although there are cases where the target image is occluded, it is a minority after all. There must be an unoccluded target in any image, so the target image with the smallest variation must be complete) to perform target type analysis (circular or annular), thereby completing the viewing direction. The specific steps of R1 are as follows: I1: Binarize the forward and backward target light images respectively, obtain the outer boundary through Canny edge detection, and obtain the spot area; restore the center coordinates of the target light.

[0311] I2: Calculate the variability E of the light spot in the image R , according to E R To determine the viewing direction.

[0312] I3: For each image, select the light spot with the smallest variation to determine the viewing direction.

[0313] I1 includes: The forward and backward target light images are binarized separately.

[0314] The identified boundary pixel center is stored as the boundary point (x Bi , j , y Bi , j ), (x Bi , j , y Bi , j ) represents the coordinates of the jth boundary point of the i-th light spot; the total pixel area A inside the boundary is calculated i .

[0315] The center point G coordinate (x Gi , j , y Gi , j ), (x Gi , j ,y Gi , j ) represents the center of the i-th spot. The central dark area is not considered when calculating the backward target light. Gi , j , y Gi , j ): .

[0316] I2 includes: I21: Calculate the centroid G of the i-th spot iThe distance S to the jth boundary point i , j : .

[0317] I22: Calculate the equivalent circle of the i-th spot (such as Figure 32 The dotted circle shows the radius R ei : .

[0318] I23: For the i-th light spot, the equivalent circle radius R ei As a reference, calculate S i , j and the equivalent circle radius R ei Deviation, E Ri That is the variability of the light spot: .

[0319] E Ri The smaller the value, the less the contour change caused by occlusion. When the light spot is a standard circle, the value should be 0.

[0320] I3 includes: I31: Taking the center of the target light spot as the starting point and the outer boundary of the detection as the limit, calculate the distribution of the gray value k of the image along the x and y directions respectively, and calculate the corresponding gray sample variance S respectively x and S y (like Figure 33 shown): .

[0321] .

[0322] I32: Construct the target discriminant: .

[0323] I33: If Δ=0, the target light in the image is determined to be forward-facing; if Δ≠0, the target light in the image is determined to be backward-facing.

[0324] The above step R1 further includes: According to the variation E R Combined spot area A i The integrity of the light spot is evaluated and the light spots that meet the following conditions should be eliminated: .

[0325] in is the variation threshold of the unobstructed light spot.

[0326] Variability calibration primarily considers two aspects: For images of the same wavelength, the variation is related to the spot size (i.e., the distance of the target light from the camera). Since the image resolution is fixed, the smaller the spot size, the more jagged the edge, and the greater the resulting variability. Furthermore, sensors with different wavelengths have different imaging areas, requiring calibration of the variation threshold for each wavelength. The variability calibration process includes the following steps: Calibrate the variability threshold and determine the initial value of the variability threshold: Q1: Use the light targets arranged on the site as calibration objects, and set all light targets to λ through the remote host 600. k Wavelength, k≤b, where b is the total number of wavelengths that each target light can emit, and all of them are lit.

[0327] Q2: Shooting λ k 1 image at a wavelength, and obtain the area A of each light spot in the image k,i , spot variability E Rk,i ; i≤n, n is the total number of light spots in the image (at this time, there may be afterimages in each light spot, and not all light spots are complete).

[0328] Q3: Keep E Rk,i With A k,i The corresponding relationship between A k,i Arrange into sequence k , n}, and obtain the corresponding spot variation sequence {E Rk,n}.

[0329] Q4: Find the sequence {E Rk,n}'s minimum value term E Rk,m , sequence {A k , n The corresponding item is A k , m .

[0330] Q5: Find the sequence {E Rk,n}'s maximum value term E Rk,p , sequence {A k , n The corresponding item is A k , p .

[0331] Q6: If A in Q5 k , p In the sequence {A k , n} is the smallest, then calculate λ as follows k The area under the wavelength is A k,i The spot variation threshold E R0(k，i) : .

[0332] Q7: If A in Q5 k , p In the sequence {A k , n} is not the smallest, remove the item and re-execute Q5.

[0333] To facilitate understanding of the above process, an example is given below: The image at a certain wavelength is calculated to obtain the area and variation of each spot as follows: As can be seen from the above table, among them: Minimum variation E Rk,2 =1.802, which corresponds to A k , 2=270; Maximum variation E Rk,3 =3.520, which corresponds to A k , 3=151 <min({ A k , n})=145, remove and screen again; Maximum variation E Rk,3 =2.102, which corresponds to A k , 4=145=min({ A k , n})=145, meets the requirements; When the spot area is known to be 255, its variation threshold E R0 =1.838.

[0334] The advantage of this method for calculating the variability threshold is that it does not require the deployment of separate targets off-site for calibration; calibration can be achieved directly using targets deployed on-site. In addition, the variability threshold is adapted to the wavelength and spot area, taking into account the influence of area differences in different bands and different imaging areas, resulting in higher judgment accuracy.

[0335] In the above step R2, the light spots from the same target light in the images of each wavelength are merged, mainly to solve the problem of combining and matching the images of the target lights of each wavelength. After the target light is processed by the above-mentioned residual image removal, the light spots from the light spots of the images of each wavelength are screened out to determine which ones come from the same target light, so that the different wavelength images at the same position can be correctly combined together, and the center of the target light after the multi-band image combination is solved.

[0336] Since each imaging element of the multi-wavelength imaging camera has the same image plane coordinate system, the multi-band images will be superimposed when the images of each band are projected onto the virtual imaging surface (the virtual imaging surface is only imagined for the sake of ease of understanding, and the light of each wavelength will not overlap and form a secondary image). Figure 34 As shown in the figure, taking three wavelength target lights as an example, when all three wavelength lamp beads are lit, the three images will be superimposed on the virtual imaging surface, and the same applies to the remaining lamp bead combinations. To achieve the above superposition process, it is necessary to first establish a mapping relationship for the spot data of each band image, such as Figure 35 As shown, the specific steps include: J1: Obtain the outer boundary and center coordinates of the target light image of each wavelength, and count the number of light spots in the image of each wavelength respectively. The central dark area of the backward target light is not considered in the analysis.

[0337] J2: For target lights with b wavelengths, the wavelength image with the largest number of light spots is used as the benchmark, and its wavelength is λ t , calculate the distance between the center of each spot image of this wavelength and the center of the spot of other wavelength images: .

[0338] Among them, x k The horizontal coordinate of the center of the kth spot in the wavelength image with the largest number of spots; y k Indicates the vertical coordinate of the center of the kth spot in the wavelength image with the largest number of spots; x λc,n Indicates wavelength is λ c The horizontal coordinate of the center of the nth spot in the wavelength image; λc,n Indicates wavelength is λ c The vertical coordinate of the center of the nth spot in the wavelength image; D(k, λ c , n) represents the kth light spot in the wavelength image with the largest number of light spots and the wavelength λ c The center distance of the nth light spot in the wavelength image.

[0339] J3: At wavelength λ c Among the many points, when a certain point can make the above-calculated D(k, λ c , n) value is the smallest, then the point has the largest number of spots and the wavelength is λ t The k-th point distance is used to achieve overlap; using D(k, λ c, n), the wavelength is established as λ according to the following formula t The corresponding relationship between each target point and other wavelength targets: .

[0340] Among them, P(λ t , k) indicates the largest number of light spots with a wavelength of λ t The kth point of P(λ c , q) represents the wavelength λ c The qth point of P(λ t , k) overlap.

[0341] J4: Merge the light spot centers of the same target lights.

[0342] The merging process from J1 to J3 mentioned above only completes the classification of the spot images of different wavelengths. In the actual imaging process, due to factors such as the difference in the sensitivity of the sensor to different wavelengths, for the same target light, it is impossible to achieve complete overlap in spatial position and shape of images of different wavelengths. There will always be deviations in position and shape (such as Figure 36 -(a)). However, when using the multi-band images of the target light to overlap and determine the wavelength combination and spot center of the target light, we hope that the centers of the light spots of each wavelength are coincident. Therefore, it is necessary to merge the spot centers of the same target light determined in J3. For the same target light with three wavelengths, the merging method in J3 is used. There are at most three image merging categories, namely: the first category, single wavelength, no merging is required; the second category, two-wavelength image merging; the third category, three-wavelength image merging. However, the so-called merging is not simply to overlap these points, but to construct an algorithm to merge the points equivalently into the optimal point. The characteristic of the merged optimal point should be that the sum of the distances between the optimal point and each known point is the smallest. Based on this principle, the specific merging method in step J4 is as follows: J41. Equivalently merge the spot centers of different wavelengths of the same target light.

[0343] J42. Merge the spot areas of different wavelengths of the same target light.

[0344] J41 includes: J411: Set the resolution accuracy: When calculating according to J412 to J418, as the search center gradually approaches the search center point M, the step size will tend to 0; considering the discreteness of digital imaging, Δ≤δ is used as the cutoff condition, where Δ is the step size and δ is the pixel size of the imaging sensor.

[0345] J412: Determine the initial search area: Based on the center point P of the light spot of the same target light with different wavelengths λ1 ~P λm(1≤m≤b) coordinates determine the initial search range, take min(x i )≤x≤max(x i ), min(y i )≤y≤max(y i ) (1≤i≤m) area as the initial area; J413: Determine the initial search center: J412 determines the geometric center G formed by all points in the area as the initial search center point M, the coordinates are (such as Figure 36 -(a)): .

[0346] J414: Determine the initial search step size: half of the minimum value of the height and width direction of the area determined by J412 is used as the initial search step size Δ (such as Figure 36 -(b)): .

[0347] J415: Search area: With the optimal point M as the center and Δ as the step size, search points are arranged in 8 directions orthogonal to its periphery. For the first calculation, point M is point G; the coordinates of each search point are expressed as (e.g. Figure 36 -(b)): .

[0348] J416: Target calculation: Calculate the distance between each search point arranged in J415 and the center points of m light spots of the same target light with different wavelengths and sum them up (e.g. Figure 36 -(c)): .

[0349] J417: According to J411, the accuracy is judged. If it meets the requirements, the point M is output, which is the optimal point. Otherwise, the search center and step size are adjusted according to J418 (such as Figure 36 -(d)); J418: For point M that does not meet the conditions of J417, consider the following two situations: Case 1: If M coincides with the center point M in J415, it indicates that the initial search area is too large. In this case, update the step size Δ in step J415 to Δ / 2 to further narrow the range; then recalculate according to steps J416 to J417 (e.g. Figure 36 -(d)).

[0350] Case 2: If M does not coincide with the center point M in J415, the position of point M should be updated. At this time, point M in step J415 is updated to this point, and then steps J416 to J417 are executed (as shown in the following example). Figure 36 -(e)).

[0351] J419: Follow steps J411 to J418 to point P λ1 ~P λm Processing is performed to complete the merging of the centers of light spots of different wavelengths of the same target light.

[0352] In the above steps, we have completed the equivalent merging of the centers of the light spots of different wavelengths of the same target light into one point (i.e., position merging). Next, we continue to merge the areas of the light spots of different wavelengths of the same target light (shape merging), thus completing the superposition of multiple band images. Multi-band image area merging is not simply adding up the areas of each image and calculating the average. The weight of the position distribution deviation should be considered and the weighted average of each area should be calculated. Based on this principle, in step J4, the specific merging method step J42 includes: J421: For the light spots of the same target light with different wavelengths, calculate their merged center M (x M , y M ), and then calculate the square of the distance S between the merge center M and the center of each wavelength spot i : ; J422: Based on the minimum distance deviation D calculated in J416, calculate the distance deviation weight β between the spot center of the same target light with different wavelengths and the merge center M: ; J423: According to the spot area A of the same target light with different wavelengths i , considering the image of J422 with medium-range deviation weight, calculate the weighted mean of the superposition of the spot areas of each wavelength: ; J424: Follow steps J421 to J423 to point P λ1 ~P λm The corresponding spot areas are processed to complete the merging of the spot areas of different wavelengths of the same target light.

[0353] The above has completed the target image recognition, residual image elimination, and homologous target image merging, but these are only for the image of one target light. There are several target lights on a target, so it is necessary to classify the many target images processed above and classify the target light images on the same target to determine the target's spatial position. Since the target lights on the same target are in the same plane, the distance between each target light and the camera is very small, so the aberration of their respective spot areas is very small; but because the scale variation of the spatial distribution of different targets is much greater than the distance between each target light and the camera, the images can be classified based on area, and those with similar areas are classified as the same target. Based on this principle, the specific target light clustering method in step R3 is as follows: R31: Suppose there are q homologous target lights merged data, and the merge center of each homologous target image is (x Mi , y Mi ), the merged area is A Ci (i≤q), the central dark area of the rearward target light is not considered in the analysis; R32: A Ci Rearrange in ascending order; R33: From A C1 To start, press A Ci With A Ci+1 Compare and calculate the relative deviation value δ i : ; R34: If δ i >δ0, then A Ci+1 With A Ci Targets classified into the same category are recorded as category r; Otherwise, A Ci+1 The target markers classified into the next category are classified as category r+1; Where δ0 is the area resolution; R35: According to the above steps R31 to R34, if there are t types of targets in total, then discard the last type of grouped targets, that is, discard the tth group, and retain the first t-1 groups as the target clustering result (such as Figure 37 As shown in the figure, the ninth category cannot be distinguished from its area change, so only targets from categories 1 to 8 are taken, and the ninth category is discarded because it cannot be distinguished from its target).

[0354] The aforementioned area resolution δ0 is a characteristic of the light source itself and needs to be calibrated to determine its initial value. The specific calibration method is as follows: Calibrate the area resolution and determine the initial value of the area resolution: P1: Use the light targets arranged on the site as calibration objects, and set all light targets to λ through the remote host 600. k Wavelength, k≤b, where b is the total number of wavelengths that each target light can emit, and all of them are lit.

[0355] P2: Shooting λ k An image under a wavelength, obtain the area A of each light spot in the image k,i , i≤n, n is the total number of light spots in the image.

[0356] P3: A k,i Sort in descending order to get the sequence {A k , n}, n is the total number of spots in the image, and the difference between adjacent items Δ is calculated k,j =A k , j+1 -A k, j ; j≤n, n is the total number of light spots in the image (at this time, there may be afterimages in each light spot, and not all light spots are complete).

[0357] P4: Take δ k , 0= min(Δ k,j ) and δ k ,0>0 is taken as the area resolution of the image at this wavelength.

[0358] P5: Calculate the δ of each spot image under type b wavelength according to P1 to P4 k ,0 value.

[0359] P6: Calculate the variation value of each spot image under the wavelength of type b respectively, and select the spot with the smallest variation value as the calculation object.

[0360] P7: Take the light spot with the smallest wavelength variation value in each category as the calculation object, and calculate the deviation weight β corresponding to each wavelength of the light spot. k .

[0361] P8: The δ calculated in P5 k ,0 value is calculated according to β in P7 k Perform weighted averaging to obtain the area resolution δ0: .

[0362] The advantage of this method in calculating and determining the area resolution threshold is that there is no need to set up separate targets off-site for calibration, and calibration can be achieved directly using the targets arranged on site. Because the merged area of the same-source but different-wavelength target images is used for spot clustering, this resolution threshold takes into account the influence of the area differences of imaging in different bands, and the judgment accuracy is higher.

[0363] After completing the above decoding steps, the method of this embodiment further includes (refer to Figure 38 ): Integrating the decoding results of the target light image set into image labels to obtain first data specifically includes the following steps: V1: Record the timestamp and camera number when the image is taken.

[0364] V2: Determine the viewing direction of the image and the available target images for screening, using code 1 to represent forward direction and 0 to represent backward direction.

[0365] V3: Different target lights can be distinguished based on wavelength combination.

[0366] V4: Complete the coordinates of each target light image in the image plane, and use (x i , y i )Record.

[0367] V5: Find all the target lights contained in each target and store the positions of the target lights in the same group as a set.

[0368] V6: Generate a coding sequence using the set in V5 so that each target light image has a unique identifiable number.

[0369] V6 includes: V61: If V2 is judged as 1, then for each set in V5, sort the x coordinates of the target lights contained therein in ascending order; for example, if a target light set is {(12, 2), (5, 5), (8, 9)}, then x The coordinates are arranged in ascending order, and the sequence after ascending order is: (5, 5), (8, 9), (12, 2).

[0370] If V2 is judged as 0, then for each set in V5, sort the x coordinates of the target lights contained therein in descending order; for example, if a target light set is {(12, 2), (5, 5), (8, 9)}, sort them by x The coordinates are arranged in descending order, and the sequence after descending order is: (12, 2), (8, 9), (5, 5).

[0371] V62: Replace the coordinates of the target light sequence in V61 with the corresponding target light code in V3 to obtain a new sequence, which is the entire target number. For example, if the target light sequence is (12, 2), (8, 9), (5, 5), and the corresponding target light codes are 001, 110, 101, then the target number is 001110101.

[0372] The above method associates and integrates the target light direction, target light image coordinates, and target number information without the help of any reference other than a single image, thus realizing the autonomous identification of the target image.

[0373] Further, refer to Figure 39 As shown, to complete the target autonomous identification, the following steps are also included: N1: When the camera is triggered to capture images, each wavelength of the target light image will generate a single-channel image file in the camera and transmit it to the remote host 600 through intra-field communication.

[0374] N2: The remote host 600 calculates and stores the image label.

[0375] N3: Generate visual overlay images: N4: The remote host 600 stores the camera's "Unix timestamp", "camera number" and the "viewing direction", "band number", "target image code" and "target image center" obtained in N2 into a data file DATA.

[0376] N5: Integrate the images of each band, the visual overlay image, and the data file DATA into one TIFF file.

[0377] Wherein, step N3 includes: N31: Generate a blank image with the same resolution as the images in each band and the color level of each pixel in each channel is 0.

[0378] N32: Traverse each band image according to the following rules and assign a value to each pixel level of the blank image generated by N31: For any pixel (i, j) in the image plane coordinate system of each band image, if the color level value affected by any of the images is 255, the color level of the pixel (i, j) at the corresponding position in the blank image generated by N31 is assigned a value of 255; otherwise, it is assigned a value of 0.

[0379] To make it easier to understand the above process, let's take an example. For example, the label of an image is: (1694617868,01,100110001,100110010,100110011,491.880,514.255,906.287,675.024,1320.694,835.793,2548.923,2550.323,2806.690,2456.659,3064.456,2362.994) The annotations are as follows: Unix timestamp: 1694617868 It indicates that the image was taken at 23:11:08 on September 13, 2023.

[0380] View direction: 01 Indicates that the current viewing direction is forward.

[0381] Target image code: 100110001, 100110010, 100110011 Indicates that the image contains three targets, 233, 234, and 235, and the three lights of target 233 are λ 1. λ 1+ λ 2. λ 3; The three lights of the 234 target are: λ 1. λ 1+ λ 2. λ 2; The three lights of the 235 target are: λ 1. λ 1+ λ 2. λ 2+ λ 3.

[0382] Target image center: 491.880,514.255 ,906.287,675.024, 1320.694,835.793 ,2548.923,2550.323, 2806.690,2456.659 ,3064.456,2362.994 Indicates 233 targets λ 1. λ 1+ λ 2. λ 3 The corresponding center coordinates of the three target light images are (491.880, 514.255), (906.287, 675.024), and (1320.694, 835.793).

[0383] The corresponding center coordinates of the three target light images of 234 targets λ1, λ1+λ2, and λ2 are (2548.923, 2550.323), (2806.690, 2456.659), and (3064.456, 2362.994), respectively.

[0384] Through a TIFF file, the images of each band and the data information after joint processing are integrated. This data processing method can effectively solve the problem of a large number of images collected by each camera and the difficulty of storage and classification. Moreover, since it contains all the information starting from the original image, it has strong traceability and is convenient for later verification and proofreading.

[0385] Example of a spatial multi-target image autonomous identification system: A spatial multi-target image autonomous identification system, comprising: The targets as above, the imaging devices as above, and the encoding devices as above; the number of targets arranged is a first preset number, and are staggered along a preset baseline, and the target lights on all targets have the same shape and the same display direction of light; the target lights of adjacent targets on the same baseline are staggered in the shooting direction of the imaging device; the number of imaging devices arranged is a second preset number, and are used to collect images of target lights on the targets; the encoding device is used to set the on and off combination of the target lights, encode the targets, and decode the target lights images collected by the imaging device to complete autonomous identification.

[0386] Example of a method for autonomously identifying multiple spatial target images: A spatial multi-target image autonomous identification method, based on the above-mentioned spatial multi-target image autonomous identification system, includes: U1. Arrange target clusters: Arrange a first preset number of targets in a staggered manner along a preset baseline. The target lights on all targets have the same shape and display direction of light; the target lights of adjacent targets on the same baseline are staggered in the shooting direction of the imaging device.

[0387] U2. Target coding: The target is coded by setting the on and off combination of the target lights on each target through the coding device.

[0388] U3. Capture and decode target light images: Use an imaging device to capture target light images on the target to obtain a target light image set; decode the target light image set to obtain decoded data, which includes target light direction, target light image coordinates, and target number information to which the target light belongs.

[0389] U4, autonomous identification: associate the decoded data with the target light image set to complete autonomous identification.

[0390] In U1, there are two or more preset baselines, and the preset baselines are continuous spatial curves on any wall in the application scene.

[0391] Although the present invention has been disclosed above with reference to preferred embodiments, this is not intended to limit the present invention. Any person skilled in the art can, without departing from the scope of the technical solution of the present invention, utilize the technical content disclosed above to make many possible changes and modifications to the technical solution of the present invention, or modify it into an equivalent embodiment with equivalent changes. Therefore, any simple modification, equivalent change, and modification made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall fall within the scope of protection of the technical solution of the present invention.

Claims

1. A target encoding method, characterized in that: include: set up There are a target lamps (200) on a single target, each target lamp (200) has lamp beads (204) of b wavelengths, and excluding the target where the target lamp (200) is completely extinguished, each target lamp (200) has a total of 2 b -1 combination; and each target has a target lamp (200), so for a target with a target lamp (200) and each target lamp (200) has b wavelengths, the number of target numbers that can be expressed is: (2 b -1) a , the encoding method is: F1: Number of bits for encoding a single target: Each target light (200) has lamp beads (204) of b wavelengths, and each wavelength of the lamp bead (204) requires 1 bit of encoding to represent two states of on or off, where 1 represents on and 0 represents off. Therefore, each target light (200) requires b bits of encoding, and a target light (200) of a single target requires a·b bits of encoding; F2: Single target light coding: For any target light i on the target, the on / off status codes of b wavelengths are sequentially connected to form a target light coding to represent the identity of the current target light (200), and each target light coding has b bits; F3: Single target code: Connect the single target light codes of the first target light (200) to the bth target light (200) in order from left to right to obtain the code of the entire target; F4: Convert target number and target code.

2. The target encoding method according to claim 1, characterized in that The F4 includes: Convert target number to target code: G1: Assume that the user inputs the decimal number of the target to be coded via the handheld terminal (601) or the remote host (600) as k 10 , compare A1 to A7 with the existing code. If it is the same as the existing code, return to prompt for re-entry, otherwise enter G2; G2: Assume that each target has a target lights (200), each target light (200) has b wavelengths, and the decimal number K is calculated by the first calculation unit (6013) or the remote host (600). 10 Convert to 2 b Base code K 2b , where number K 10 The initial value of is: ； G3: Check K through the first computing unit (6013) or the remote host (600) 2b Does it contain 0? If it does, then K 10 After adding 1, return to G2 and recalculate; if it does not contain 0, enter G4; G4: Calculate K 10 -K0+1, and the decimal number k entered by the user 10 Compare, if K 10 -K0+1=k 10 , then K 10 Convert to binary code K2 and write into the second storage unit (6022) according to the method from A1 to A7; if K 10 -K0+1≠k 10 , then K 10 After adding 1, return to G2 and recalculate.

3. The target encoding method according to claim 2, characterized in that: The F4 also includes: Convert target code to target number: H1: According to B1 to B3, the binary code K2 is read from the second storage unit (6022) by the handheld terminal (601) or the remote host (600); H2: Convert the binary code K2 to 2 through the first computing unit (6013) or the remote host (600) b Base code K 2-2b ; H3: The decimal code K is converted by the first computing unit (6013) or the remote host (600) 10 Convert to 2 b Base code K 2b , where K 10 The initial value K0 is calculated in the same way as G2; H4: K is calculated by the first computing unit (6013) or the remote host (600) 2-2b With K 2b For comparison, if K 2b =K 2-2b , then K 10 -K0+1 output is the decimal number of binary code K2; if K 2b ≠K 2-2b , then K 10 After adding 1, return to H3 and recalculate.

4. A method for decoding a target light image, which decodes a target light image set obtained based on the target encoding method according to any one of claims 1 to 3, characterized in that: The following steps are involved: R1. Identify the target light spots and target light directions on the images of each wavelength in the target light image set, and remove the target light spots that do not meet the preset requirements; R2, screening the light spots of each wavelength image, merging the light spots from the same target light (200) in each wavelength image to obtain a target light image, and solving the coordinates and area of the center of the target light image; R3. Classify the target lights (200) according to the area of the target light images, and classify the target lights (200) with the same area as the same target.

5. The method for decoding target light images according to claim 4, characterized in that: The R1 includes: I1: Binarize the forward and backward target light images respectively, obtain the outer boundary through Canny edge detection, and obtain the spot area; restore the target light center coordinates; I2: Calculate the variability E of the light spot in the image R , according to the E R Determine viewing direction; I3: For each image, select the light spot with the smallest variation to determine the viewing direction.

6. The method for decoding target light images according to claim 5, characterized in that: The I1 includes: Perform binarization processing on the forward and backward target light images respectively; The identified boundary pixel center is stored as the boundary point (x Bi , j , y Bi , j ), (x Bi , j , y Bi , j ) represents the coordinates of the jth boundary point of the i-th light spot; the total pixel area A inside the boundary is calculated i ; The center point G coordinate (x Gi , j , y Gi , j ), (x Gi , j , y Gi , j ) represents the center of the i-th spot. The central dark area is not considered when calculating the backward target light. Gi , j , y Gi , j ): 。 7. The method for decoding target light images according to claim 6, characterized in that: The I2 includes: I21: Calculate the centroid G of the i-th spot i The distance S to the jth boundary point i , j : ； I22: Calculate the equivalent circle radius R of the i-th spot ei : ； I23: For the i-th light spot, the equivalent circle radius R ei As a reference, calculate S i , j and the equivalent circle radius R ei Deviation, E Ri That is the variability of the light spot: ； E Ri The smaller the value, the less the contour change caused by occlusion. When the light spot is a standard circle, the value should be 0.

8. The method for decoding target light images according to claim 7, characterized in that: The I3 includes: I31: Taking the center of the target light spot as the starting point and the outer boundary of the detection as the limit, calculate the distribution of the gray value k of the image along the x and y directions respectively, and calculate the corresponding gray sample variance S respectively x and S y : ； ； I32: Construct the target discriminant: ； I33: If Δ=0, it is determined that the target light (200) in the image is facing forward; if Δ≠0, it is determined that the target light (200) in the image is facing backward.

9. The method for decoding target light images according to claim 8, characterized in that: The R1 further comprises: According to the variation E R Combined spot area A i The integrity of the light spot is evaluated and the light spots that meet the following conditions should be eliminated: ； in is the variation threshold of the unobstructed light spot.

10. The method for decoding target light images according to claim 9, characterized in that: Also includes: Calibrate the variation threshold and determine the initial value of the variation threshold: Q1: Use the light targets arranged on the site as calibration objects, and set all light targets to λ through the remote host (600). k Wavelength, k≤b, where b is the total number of wavelengths that each target light (200) can emit, and all are lit; Q2: Shooting λ k 1 image at a wavelength, and obtain the area A of each light spot in the image k,i , spot variability E Rk,i ; i≤n, n is the total number of light spots in the image; Q3: Keep E Rk,i With A k,i The corresponding relationship between A k,i Arrange into sequence k , n }, and obtain the corresponding spot variation sequence {E Rk,n }; Q4: Find the sequence {E Rk,n }'s minimum value term E Rk,m , sequence {A k , n The corresponding item is A k , m ; Q5: Find the sequence {E Rk,n }'s maximum value term E Rk,p , sequence {A k , n The corresponding item is A k , p ; Q6: If A in Q5 k , p In the sequence {A k , n } is the smallest, then calculate λ as follows k The area under the wavelength is A k,i The spot variation threshold E R0(k，i) : ； Q7: If A in Q5 k , p In the sequence {A k , n } is not the smallest, remove the item and re-execute Q5.

11. The method for decoding target light images according to claim 10, characterized in that: The R2 comprises: J1: Obtain the outer boundary and center coordinates of the target light image at each wavelength, and count the number of light spots in the image at each wavelength. The central dark area of the backward target light is not considered in the analysis. J2: For a target light (200) with b wavelengths, the wavelength image with the largest number of light spots is used as a reference, and its wavelength is λ t , calculate the distance between the center of each spot image of this wavelength and the center of the spot of other wavelength images: ； Among them, x k The horizontal coordinate of the center of the kth spot in the wavelength image with the largest number of spots; y k Indicates the vertical coordinate of the center of the kth spot in the wavelength image with the largest number of spots; x λc,n Indicates wavelength is λ c The horizontal coordinate of the center of the nth spot in the wavelength image; λc,n Indicates wavelength is λ c The vertical coordinate of the center of the nth spot in the wavelength image; D(k, λ c , n) represents the kth light spot in the wavelength image with the largest number of light spots and the wavelength λ c The center distance of the nth light spot in the wavelength image; J3: At wavelength λ c Among the many points, when a certain point can make the above-calculated D(k, λ c , n) value is the smallest, then the point has the largest number of spots and the wavelength is λ t The k-th point distance is used to achieve overlap; using D(k, λ c , n), the wavelength is established as λ according to the following formula t The corresponding relationship between each target point and other wavelength targets: ； Among them, P(λ t , k) indicates the largest number of light spots with a wavelength of λ t The kth point of P(λ c , q) represents the wavelength λ c The qth point of P(λ t , k) overlap; J4: Merge the light spot centers of the light sources determined to be the same target light (200).

12. The method for decoding target light images according to claim 11, characterized in that: The J4 includes: J41. Equivalently merging the centers of the light spots of different wavelengths of the same target light (200); J42. Merge the spot areas of the same target light (200) with different wavelengths.

13. The method for decoding target light images according to claim 12, characterized in that: The J41 includes: J411: Set the resolution accuracy: When calculating according to J412 to J418, as the search center gradually approaches the search center point M, the step size will tend to 0. Considering the discreteness of digital imaging, Δ≤δ is used as the cutoff condition, where Δ is the step size and δ is the pixel size of the imaging sensor. J412: Determine the initial search area: Based on the center point P of the light spot with different wavelengths of the same target light (200) λ1 ~P λm The coordinates determine the initial search range, where 1≤m≤b, and min(x i )≤x≤max(x i ), min(y i )≤y≤max(y i ) area as the initial area, where 1≤i≤m; J413: Determine the initial search center: J412 determines the geometric center G formed by all points in the area as the initial search center point M, the coordinates are: ； J414: Determine the initial search step size: half of the minimum value of the height and width directions of the area determined by J412 is used as the initial search step size Δ: ； J415: Search area: With the optimal point M as the center and Δ as the step size, search points are arranged in 8 directions orthogonal to it. For the first calculation, point M is point G; the coordinates of each search point are expressed as: ； J416: Target calculation: Calculate the distance between each search point arranged in J415 and the center points of m light spots of the same target light (200) with different wavelengths and sum them up: ； J417: Perform accuracy judgment according to J411. If the accuracy is satisfied, output point M, which is the optimal point. Otherwise, adjust the search center and step size according to J418. J418: For point M that does not meet the conditions of J417, consider the following two situations: Case 1: If M coincides with the center point M in J415, it indicates that the initial search area is too large. In this case, update the step size Δ in J415 to Δ / 2 to further narrow the range; then recalculate according to J416-J417; Case 2: If M does not coincide with the center point M in J415, the position of point M should be updated. In this case, point M in J415 is updated to this point, and then J416 to J417 are executed. J419: Press J411~J418 to point P λ1 ~P λm Processing is performed to complete the merging of the centers of light spots of different wavelengths of the same target light (200).

14. The method for decoding a target light image according to claim 13, wherein: The J42 includes: J421: For the light spots of different wavelengths of the same target light (200), calculate their merged center M (x M , y M ), and then calculate the square of the distance S between the merge center M and the center of each wavelength spot i : ； J422: Based on the minimum distance deviation D calculated in J416, calculate the distance deviation weight β between the spot center of the same target light (200) with different wavelengths and the merge center M: ； J423: According to the spot area A of the same target light (200) with different wavelengths i , considering the image of J422 with medium-range deviation weight, calculate the weighted mean of the superposition of the spot areas of each wavelength: ； J424: Press J421~J423 to point P λ1 ~P λm The corresponding light spot areas are processed to complete the merging of light spot areas of different wavelengths of the same target lamp (200).

15. The method for decoding target light images according to claim 14, characterized in that: The R3 further comprises: R31: Suppose there are q homologous target lights merged data, and the merge center of each homologous target image is (x Mi , y Mi ), the merged area is A Ci , i≤q, the central dark area of the rearward target light is not considered in the analysis; R32: A Ci Rearrange in ascending order; R33: From A C1 To start, press A Ci With A Ci+1 Compare and calculate the relative deviation value δ i : ； R34: If δ i >δ0, then A Ci+1 With A Ci Targets classified into the same category are recorded as category r; Otherwise, A Ci+1 The target markers classified into the next category are classified as category r+1; Where δ0 is the area resolution; R35: According to R31 to R34 above, there are t types of targets in total. Then discard the last type of grouped targets, that is, discard the tth group, and retain the first t-1 groups as the result of target clustering grouping.

16. The method for decoding target light images according to claim 15, characterized in that: Also includes: Calibrate the area resolution and determine the initial value of the area resolution: P1: Use the light targets arranged on the site as calibration objects, and set all light targets to λ through the remote host (600) k Wavelength, k≤b, where b is the total number of wavelengths that each target light (200) can emit, and all are lit; P2: Shooting λ k An image under a wavelength, obtain the area A of each light spot in the image k,i , i≤n, n is the total number of light spots in the image; P3: A k,i Sort in descending order to get the sequence {A k , n }, n is the total number of spots in the image, and the difference between adjacent items Δ is calculated k,j =A k , j+1 -A k , j ; j≤n, n is the total number of light spots in the image; P4: Take δ k , 0= min(Δ k,j ) and δ k ,0>0 is taken as the area resolution of the image at this wavelength; P5: Calculate the δ of each spot image under type b wavelength according to P1 to P4 k ,0 value; P6: Calculate the variation value of each spot image under the wavelength of type b respectively, and select the spot with the smallest variation value as the calculation object; P7: Take the light spot with the smallest wavelength variation value in each category as the calculation object, and calculate the deviation weight β corresponding to each wavelength of the light spot. k ; P8: The δ calculated in P5 k ,0 value is calculated according to β in P7 k Perform weighted averaging to obtain the area resolution δ0: 。 17. The method for decoding a target light image according to any one of claims 4 to 16, characterized in that: Also includes: The decoding results of the target light image set are integrated into image labels to obtain first data.

18. The method for decoding target light images according to claim 17, characterized in that: Integrating the decoding results of the target light image set into image labels includes: V1: Record the timestamp and camera number when the image was taken; V2: Determine the viewing direction of the image and the available target images for screening, with code 1 representing forward direction and 0 representing backward direction; V3: Differentiate between different target lights (200) according to wavelength combination; V4: Complete the coordinates of each target light image in the image plane, and use (x i , y i )Record; V5: Find all target lights (200) contained in each target, and store the positions of target lights in the same group as a set; V6: Generate a coding sequence using the set in V5 so that each target light image has a unique identifiable number.

19. The method for decoding target light images according to claim 18, characterized in that: The V6 includes: V61: If the result in V2 is 1, then for each set in V5, sort the x-coordinates of the target lights contained therein in ascending order; If V2 is judged to be 0, then for each set in V5, sort the x-coordinates of the target lights contained therein in descending order; V62: Replace the coordinates corresponding to the target light sequence in V61 with the corresponding target light code in V3 to obtain a new sequence, which is the entire target number.

20. The method for decoding target light images according to claim 19, characterized in that: It also includes completing the target autonomous identification, the steps are as follows: N1: When the camera is triggered to capture images, each wavelength of the target light image will generate a single-channel image file in the camera and transmit it to the remote host through intra-field communication (600); N2: The remote host (600) calculates and stores the image tag; N3: Generate visual overlay images: N4: The remote host (600) stores the camera's "Unix timestamp", "camera number" and the "viewing direction", "band number", "target image code" and "target image center" obtained in N2 into a data file DATA; N5: Integrate the images of each band, the visual overlay image, and the data file DATA into one TIFF file.

21. The method for decoding target light images according to claim 20, characterized in that: The N3 includes: N31: Generate a blank image with the same resolution as the images of each band and the color level of each pixel in each channel is 0; N32: Traverse each band image according to the following rules and assign a value to each pixel level of the blank image generated by N31: For any pixel (i, j) in the image plane coordinate system of each band image, if the color level value affected by any of the images is 255, the color level of the pixel (i, j) at the corresponding position in the blank image generated by N31 is assigned a value of 255; otherwise, it is assigned a value of 0.

Images