Target for Image Autonomous Identification, Target Coding Method and Image Analysis Method

Abstract

The present invention relates to the field of vision measurement, and particularly to a target for image autonomous identification, a target coding method and an image analysis method. The target includes a target rod, a target base and a target lamp. One end of the target rod is adjustably mounted on the target base, and the target lamp is installed on the target rod and can display lights on the front and back surfaces of the target rod. The target lamp includes a rear scattering cover, a rear hyperbolic reflection cavity, a reflection cone, lamp beads, a front hyperbolic reflection cavity and a front scattering cover. The rear hyperbolic reflection cavity is docked with the front hyperbolic reflection cavity and clamps and fixes the reflection cone inside. There are multiple lamp beads, which 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. The coding device includes a remote host, more than one handheld terminal and more than one target coding module. The handheld terminal is used to communicate with the remote host for database synchronization, communicate with the target coding module and perform addition, deletion, query and modification of the target coding information.

Description

Technical Field

[0001] The present invention relates to the technical field of visual measurement, and in particular to a target, a target lamp, an imaging device, a coding device and a method for autonomous identification. Background Art

[0002] When using photogrammetry technology for visual positioning, it is necessary to deploy external reference targets within the measurement range and correspond the targets on the camera image with the targets in the real space to establish a control field for the visual system. Currently, most of the reference targets remain relatively stationary with respect to the camera. By pre-identifying and marking in the images collected by the camera, the identity tags of various targets are not included in the inherent attributes of the images. Their initialization markings are mainly achieved through methods such as manual confirmation and spatial position analysis, which are applicable in small-scale scenarios where the camera is static and the number of targets is small. However, for large-scale cluster camera and multi-target scenarios (such as Figure 1 shown), the camera is in dynamic cluster operation, and it is difficult to initialize the targets for each camera one by one. It is difficult to distinguish the large number of targets identified by each camera (such as Figure 2 shown), and it is difficult to identify the targets when the position of any camera is dynamically switched (such as Figure 3 shown). In summary, it is necessary to develop a target and an autonomous identification technology that can adapt to large scenarios and meet the identification requirements of cluster dynamic cameras, and incorporate the target identity information into the image attributes to improve the accuracy and efficiency of visual positioning.

[0003] In order to achieve the above vision, it is necessary to deploy several targets with known coordinates in space. By identifying the targets on the image and assigning the corresponding known coordinates, the external elements of the camera (referring to the 3 spatial coordinates and 3 rotation postures of the camera in space) can be determined. However, there is more than one target in the image, so it is necessary to number the targets, then identify the targets on each image, and establish a one-to-one correspondence between the identified targets and the numbers.

[0004] Currently, computer graphics technology can identify targets from background images, but it is difficult to correspond the identified targets with their numbers (i.e., the initialization assignment of target coordinates), and it is often completed manually. When the number of images spanned by the targets is large, this method is inefficient. When there is an immediate calculation requirement, the images captured by the camera are often binary images with almost no background reference, and it will be even more difficult to correspond the target numbers at this time.

[0005] On the basis of realizing that the target identity tag is included in the intrinsic attributes of the image, it is necessary for the camera to capture light sources of different wavelengths emitted from the same spatial position simultaneously at the moment of shooting, and generate several images without parallax, and obtain the target number information through image superposition. This requires the camera to obtain images of the same target at the same moment, different wavelengths, equal intensity, and without parallax. Images of different wavelengths can be solved by optical filters, which are relatively mature at present; but for obtaining images of the same object, at the same moment, equal intensity, and without parallax, the main methods currently used are as follows:

[0006] Method 1: Use a multi-lens multi-filter camera

[0007] This method can achieve simultaneous capture of light sources of different wavelengths emitted from the same target. However, since multiple cameras are used for simultaneous viewfinder, the spatial positions where each camera is distributed cannot be made exactly the same (as shown in Figure 4 ), and the parallax of the captured images cannot be eliminated (as shown in Figure 5 ). After the multi-band images are stacked, the images of the target cannot coincide, and it is difficult to achieve the self-identification of the target.

[0008] Method 2: Use a single-lens multi-filter camera

[0009] This method can capture light sources of different wavelengths emitted from the same target at different times by installing a device for automatically switching filters inside the lens (as shown in Figure 6 ). Since the filter switching takes time and the camera needs to take an image after the filter switching is completed, during the process of several filter switches, the target or the camera may have undergone slight disturbances. Therefore, the images of various wavelengths captured by it do not record the position of the target at the same moment. Due to the disturbances of the camera or the target, the images of the target cannot coincide, and it is also difficult to achieve the self-identification of the target (as shown in Figure 7 ).

[0010] Method 3: Use an optical image splitting device

[0011] To achieve the imaging function of the aforementioned target, the key lies in the image splitting of the image. Currently, it mainly relies on optical methods. By using a beam splitter, the same light beam can be decomposed for imaging, but there are the following limitations: First, the current image splitting by the beam splitter is only a simple image splitting and does not have the function of filtering multi-band light sources, making it difficult to achieve synchronous imaging of multi-wave light sources at the same position. Second, no device and method for strictly controlling image parallax have been proposed, making it difficult to ensure the synchronous focusing and equal size of multiple images, resulting in parallax or defocus between images. Third, when the positions of the camera and the light source are not specific (usually a horizontal light beam in the laboratory, and there are multiple divergent light beams in reality), for the light rays split by the beam splitter, the incident angles of each light beam are different. After multiple refractions and reflections, it is impossible to ensure that the light intensities of each split image are consistent, resulting in overexposure or underexposure of the images. Due to the above three limitations, the generated images have problems such as missing details and parallax (as shown in Figure 8 ), which will cause the images to not overlap when multiple-band images are superimposed, making it difficult to achieve the self-identification of the target. Summary of the Invention

[0012] The present invention provides a target, a target lamp, an imaging device, a coding device and a method for autonomous identification, aiming to solve the technical problem of the large difficulty in initializing and identifying a large number of cameras one by one for targets in the existing vision measurement technology.

[0013] To achieve the above object, the present invention provides a target for autonomous identification of images, including a target rod, a target base and a target lamp. One end of the target rod is adjustably installed on the target base, and the target lamp is installed on the target rod and can display lights on the front and back surfaces of the target rod.

[0014] As a further improvement of the above technical solution:

[0015] The target base is provided with a chassis. A set of installation and fixing components are respectively provided at both ends of the chassis. The installation and fixing components include fixing fasteners and a pair of leveling bolts. The target base is provided with installation through holes for installing the fixing fasteners, and leveling screw holes for installing a pair of leveling bolts are respectively arranged on both sides of the installation through holes.

[0016] The bottom of the leveling bolt is spherical.

[0017] An arc guide groove is provided on the chassis. The target rod is hinged to the target base, and the hinge point is located at the center of the arc guide groove. A locking screw is slidably arranged on the arc guide groove, and the locking screw cooperates with the target rod to fix the target rod at a specified position.

[0018] There are three target lamps. The three target lamps are evenly spaced along the target rod, and the lights displayed by the three target lamps on the front and back surfaces are the same in the same direction and different in the opposite direction.

[0019] The surface of the target rod is provided with a low-reflectivity matte coating.

[0020] A target lamp for a target used for image autonomous identification as described above, comprising a backscattering cover, a back hyperbolic reflection cavity, a reflection cone, lamp beads, a front hyperbolic reflection cavity and a front scattering cover. The back hyperbolic reflection cavity is docked with the front hyperbolic reflection cavity and clamps and fixes the reflection cone inside. There are multiple lamp beads, which are dispersedly arranged on the front and back sides of the reflection cone. The front scattering cover is installed at the front end of the front hyperbolic reflection cavity, and the back scattering cover is installed at the back end of the back hyperbolic reflection cavity.

[0021] As a further improvement of the above technical solution:

[0022] A light-shielding hole is opened in the center of the front scattering cover or the back scattering cover, and a light-shielding cylinder is provided at one end of the reflection cone. The light-shielding cylinder is sleeved in the light-shielding hole.

[0023] The lamp beads include LED lamp beads of multiple different wavelengths. LED lamp beads of multiple different wavelengths are provided on both the front and back sides of the reflection cone, and the multiple different wavelengths of LED lamp beads located on the same side of the reflection cone are arranged staggeredly.

[0024] The lamp beads on the front and back sides of the reflection cone are arranged staggeredly.

[0025] The optical centers of the lamp beads on the front and back sides of the reflection cone are located on the same plane.

[0026] The back hyperbolic reflection cavity, the front hyperbolic reflection cavity and the reflection cone are coaxially arranged.

[0027] A fixing method for fixing the target for image autonomous identification as described above on an uneven surface, comprising the following steps:

[0028] 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 the fixing fastener not fully compress the target base.

[0029] S2: Leveling. Screw the leveling bolts into the respective leveling screw holes, make the spherical parts at the bottoms of the leveling bolts contact and abut against the uneven surface, and adjust the screwing depth of each leveling bolt to make the attitude of the target base tend to the target installation attitude.

[0030] S3: Fixing. When the target base reaches the target installation attitude, tighten the fixing fastener to completely fix the target base, and complete the fixed installation of the target on the uneven surface.

[0031] A device for iso-intensity parallax-free synchronous image splitting and multi-band imaging comprises a lens and a frame, wherein the lens is mounted on the frame, and a light splitting channel and a plurality of imaging components are arranged in the frame, wherein the light splitting channel transmits iso-intensity light beams formed by multi-level light splitting and reflection of light incident from the lens 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.

[0032] As a further improvement of the above technical solution:

[0033] The imaging components are provided with four groups, and the splitting channel is provided with a primary beam splitter and two secondary beam splitters. After the incident light is dispersed into two beams by the primary beam splitter, the two beams are dispersed into two beams again by a secondary beam splitter respectively, forming a total of four beams of light that are transmitted to the four groups of imaging components respectively.

[0034] 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, dispersed into two beams of light by the second-level beam splitter, and then respectively reflected into a corresponding imaging component by a second reflector and a third reflector; the second beam of light is directly directed to another second-level beam splitter, dispersed into two beams of light by the second-level beam splitter, one of which is reflected into a corresponding imaging component by a fourth reflector and a fifth reflector in sequence, and the other is reflected into a corresponding imaging component by a sixth reflector and a seventh reflector in sequence.

[0035] A secondary beam splitter and a residual light absorbing cavity are arranged between the secondary beam splitter and the imaging component.

[0036] The primary beam splitter, the secondary beam splitter and the secondary beam splitter are arranged in parallel.

[0037] 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.

[0038] A heat insulation cavity is arranged inside the outer shell surface of the frame.

[0039] A heat insulation cavity is arranged in the middle of the frame.

[0040] A coding device for target coding comprises a remote host, one or more handheld terminals and one or more target coding modules.

[0041] 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.

[0042] The target encoding module is integrated in the target and is used to synchronize the real-time encoding information of the target to the remote host; it is also used to communicate with the handheld terminal to receive add, delete, query, and modify instructions from the handheld terminal; it is further used to send the encoding and status information of the target to the handheld terminal.

[0043] As a further improvement of the above technical solution:

[0044] The handheld terminal includes a first radio frequency unit, a first storage unit, a first computing unit, and a WIFI unit.

[0045] The first radio frequency unit includes an RFID tag and is used to communicate with the target encoding module.

[0046] The first storage unit is used to store the data of the handheld terminal.

[0047] The first computing unit is used for encoding information calculation of the handheld terminal.

[0048] The WIFI unit is used to communicate with the remote host.

[0049] The target encoding module includes a second radio frequency unit, a second storage unit, a second computing unit, and a NET unit.

[0050] The second radio frequency unit includes an RFID tag and is used to communicate with the handheld terminal.

[0051] The second storage unit is used to store the data of the target encoding module.

[0052] The second computing unit is used for encoding information calculation of the target encoding module.

[0053] The NET unit is used to communicate with the remote host.

[0054] Communication between the handheld terminal and the target encoding module is carried out through the first radio frequency unit and the second radio frequency unit.

[0055] The network relationship between the remote host and the one or more target encoding modules includes:

[0056] The remote host can perform one-way read and write operations on the data of the one or more target encoding modules.

[0057] The network relationship between the remote host and the one or more handheld terminals includes:

[0058] The remote host can perform read and write operations on the data of the one or more handheld terminals.

[0059] The one or more handheld terminals can perform read operations on the data of the remote host.

[0060] The remote host can publish information to the outside through the mobile network, and the mobile network includes the network provided by the communication service provider.

[0061] The network relationship between the handheld terminal and the target coding module includes:

[0062] The handheld terminal can perform data reading on the target coding module.

[0063] Before the handheld terminal performs data reading on the target coding module, the handheld terminal needs to communicate with the remote host first for database synchronization.

[0064] A target coding control method, based on the above-mentioned coding device for target coding, includes:

[0065] T1. Communicate with the remote host through the handheld terminal for database synchronization, communicate with the target coding module, and perform addition, deletion, query, and modification on the target coding information.

[0066] T2. Communicate with the handheld terminal through the target coding module to receive the addition, deletion, query, and modification instructions from the handheld terminal; send the coding and status information of the target to the handheld terminal through the target coding module.

[0067] As a further improvement of the above technical solution:

[0068] The T1 includes:

[0069] A1: For the target with initialized or modified coding, use the handheld terminal to synchronize the coding library with the remote host through the WIFI unit and update the coding library in the first storage unit.

[0070] A2: Input the proposed target number through the handheld terminal and perform coding through the first calculation unit.

[0071] A3: The first calculation unit reads the data in the first storage unit and compares the input and encoded target number with it. If it duplicates the existing coding, return for re-entry; if not, write the input and encoded target number into the first storage unit.

[0072] A4: Trigger the RFID tag of the first radio frequency unit to transmit to the RFID tag of the second radio frequency unit of the target.

[0073] A5: When the RFID tag of the second radio frequency unit receives the coding write instruction, trigger the read and write instruction of the second calculation unit, and the second storage unit receives and writes the new coding.

[0074] A6: The new code written into the second storage unit is synchronized to the remote host through the NET unit, and its encoding library is updated.

[0075] A7: The second computing unit outputs the new code of the second storage unit to the lamp beads corresponding to each target lamp, and controls the on / off combination of the lamp beads.

[0076] The T2 includes:

[0077] B1: When the handheld terminal reaches the sensing range of the RFID tag of the second RF unit of the target, the RFID tag of the first RF unit on the handheld terminal communicates and activates with the RFID tag of the second RF unit on the target.

[0078] B2: When the RFID tag of the second RF unit on the target receives a query instruction, it triggers the read / write function of the second computing unit to read the latest target coding information and working status information in the second storage unit.

[0079] B3: The RFID tag of the second RF unit on the target feeds back the latest target coding information and working status information to the RFID tag of the first RF unit on the handheld terminal, and then decodes it through the first computing unit and displays it on the handheld terminal.

[0080] It also includes: reading, writing, collecting, and verifying the coding information of the target through the remote host, and adjusting the on / off combination of the target lamps.

[0081] Reading, writing, collecting, and verifying the coding information of the target through the remote host includes:

[0082] C1: The remote host sends a coding information collection instruction to each target.

[0083] C2: Trigger the second computing unit of each target to execute the coding information read / write instruction, and send the coding information stored in the second storage unit to the remote host through the NET unit.

[0084] C3: The remote host verifies the collected target coding information. If there is a duplicate code situation, it corrects the duplicate code, sends a code correction instruction to the corresponding target, and sends the corrected coding information to the corresponding target.

[0085] C4: The second computing unit of the corresponding target receives the code correction instruction sent by the remote host and the corrected coding information, and writes the corrected coding information into the second storage unit for refreshing.

[0086] C5: The second computing unit controls the corresponding lamp beads of the target lamp according to the corrected coding information and adjusts its on / off combination.

[0087] It also includes:

[0088] Status monitoring: When the target lamp fails, the encoding device is used to monitor the status of the target.

[0089] Power-off recovery: When the system experiences a power-off and then power-on situation and the target lamp returns to its default state, the encoding device is used to restart the target lamp to restore the preset state of the target.

[0090] The status monitoring includes:

[0091] D1: Using optical simulation, calculate the most unfavorable light distribution with different numbers of damaged lamp beads, and comprehensively consider the light uniformity and the minimum light resolution of the camera to determine the allowable number of damaged lamp beads.

[0092] D2: The second calculation unit collects the on / off status of each lamp bead at a preset frequency and generates corresponding monitoring codes; for the lamp beads of a certain wavelength in the same target lamp, when the number of damaged lamp beads collected is the allowable number in D1, the lamp bead is still recorded as fully lit, otherwise it is recorded as fully extinguished; when the lamp bead is marked as fully extinguished, the corresponding monitoring code is written into the second storage unit.

[0093] D3: Compare the monitoring codes generated in D2 with the original codes in the second storage unit. If the comparison results are the same, it is determined that the lamp beads of this wavelength are normal and the current state is maintained; if the comparison results are different, it is determined that the lamp beads of this wavelength are damaged, and the second calculation unit sends a lighting stop instruction to turn off all the target lamps of this target.

[0094] D4: For the damaged situation determined in D3, the second storage unit sends the stored monitoring code to the remote host through the NET unit and alarms for maintenance.

[0095] D5: The remote host publishes maintenance information to the off-site terminal through the mobile network.

[0096] The power-off recovery includes:

[0097] E1: Before the power-off, the second storage unit has stored the target encoding information.

[0098] E2: The second calculation unit monitors the power-on information and performs self-check on the target lamp according to D1 to D5.

[0099] E3: For the target lamp with normal self-check, the second calculation unit outputs the target encoding in the second storage unit to the corresponding lamp beads of the target lamp to restore the on / off combination of the lamp beads.

[0100] A target encoding method, based on the above-mentioned target and target lamp, includes:

[0101] Assume there are a target lamps on a single target, and each target lamp has b kinds of wavelength lamp beads. Excluding the target with all target lamps extinguished, each target lamp has a total of 2 b-1 combination; and each target has a target lights, so for a target with a target lights and each target light having b wavelengths, the number of target numbers that can be expressed is: (2 b -1) a , and the encoding method is:

[0102] F1: Number of encoding bits for a single target: Each target light has b types of light beads with different wavelengths, and each type of light bead requires 1 bit of encoding to represent 2 states of on or off, where 1 represents on and 0 represents off. Therefore, each target light requires b bits of encoding, and the a target lights of a single target require a·b bits of encoding.

[0103] F2: Encoding for a single target light: For any one target light i on the target, connect the on / off status codes of b wavelengths in sequence to form the target light encoding to represent the identity of the current target light, and each target light encoding has b bits.

[0104] F3: Encoding for a single target: Connect the encodings of the single target lights from the first target light to the bth target light in sequence from left to right to obtain the encoding of the entire target.

[0105] F4: Convert the target number and the target encoding.

[0106] As a further improvement of the above technical solution:

[0107] The said F4 includes:

[0108] Convert the target number to the target encoding:

[0109] G1: Assume that the user inputs the decimal number k of the target to be encoded through a handheld terminal or a remote host 10 , compare it with the existing encoding according to steps A1 to A7. If it is repeated with the existing encoding, return a prompt to re-enter, otherwise enter G2.

[0110] G2: Assume that each target has a target lights and each target light has b wavelengths. Convert the decimal number K 10 to a binary encoding K b through the first calculation unit or the remote host, where the initial value of the number K 2b is: 10 .

[0111] .

[0112] G3: Check whether K 2b contains 0 through the first calculation unit or the remote host. If it contains 0, add 1 to K 10 and return to G2 for recalculation; if it does not contain 0, enter G4.

[0113] G4: Calculate K 10 - K0 Add 1 and compare it with the decimal number k input by the user 10 If K 10 -K 0 +1 = k 10 , then convert K 10 to binary code K 2 and write it into the second storage unit according to the method from A1 to A7; if K 10 -K 0 +1 ≠ k 10 , then add 1 to K 10 and return it to G2 for recalculation.

[0114] The F4 further includes:

[0115] Convert the target code into a target number:

[0116] H1: According to B1 to B3, read the binary code K in the second storage unit by the handheld terminal or the remote host 2 .

[0117] H2: Convert the binary code K 2 to binary code K b through the first calculation unit or the remote host 2-2b .

[0118] H3: Convert the decimal code K 10 to binary code K b through the first calculation unit or the remote host, where K 2b The initial value of K 10 The calculation is the same as G2.

[0119] H4: Compare K 2-2b with K 2b through the first calculation unit or the remote host. If K 2b = K 2-2b , then output K 10 -K 0 +1 as the decimal number of the binary code K 2 ; if K 2b ≠ K 2-2b , then add 1 to K 10 and return it to H3 for recalculation.

[0120] A decoding method for target lamp images, based on the equal luminous intensity and parallax-free synchronous split-image and multi-band imaging device described above, obtains a target lamp image set of a target cluster composed of the above target and the above target lamp, and decodes the target lamp image set, including the following steps:

[0121] ​R1. Identify the target light spots and target light directions on each wavelength image in the target light image set, and eliminate the target light spots that do not meet the preset requirements.

[0122] R2. Screen 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.

[0123] R3. Classify the target lights according to the area of the target light image, and classify the target lights with similar areas into the same target.

[0124] The R1 includes:

[0125] I1: Perform binarization processing on the forward and backward target light images respectively, obtain the outer boundary through Canny edge detection to get the light spot area; restore the target light center coordinates.

[0126] I2: Calculate the variability E of the light spots in the image R , and determine the viewing direction according to the E R .

[0127] I3: For each image, select the light spot with the smallest variability for viewing direction discrimination.

[0128] The I1 includes:

[0129] Perform binarization processing on the forward and backward target light images respectively.

[0130] Store the center of the identified boundary pixels at the boundary point (x Bi , j , y Bi , j ), (x Bi , j , y Bi , j ) represents the coordinate of the j-th boundary point of the i-th light spot; count the total pixel area A i inside the boundary.

[0131] Restore the coordinates of its center point G (x Gi , j , y Gi , j ) through the centroid of the pixels in the boundary region, (x Gi , j , y Gi , j ) represents the center of the i-th light spot. Do not consider the central dark area when calculating the backward target light, where (x Gi , j , y Gi , j ):

[0132] .

[0133] The I2 includes:

[0134] I21: Calculate the centroid G of the i-th light spot iDistance S from the j-th boundary point i , j :

[0135] 。

[0136] I22: Calculate the equivalent circle radius R of the i-th light spot ei :

[0137] 。

[0138] I23: For the i-th light spot, with the equivalent circle radius R ei as a reference, calculate S i , j the deviation from the equivalent circle radius R ei E Ri is the variability of the light spot:

[0139] 。

[0140] E Ri The smaller the value, the lower the degree of contour change caused by occlusion. When the light spot is a standard circle, this value should be 0.

[0141] The above I3 includes:

[0142] I31: Starting from the center of the target light spot and limited by the detected outer boundary, along the x and y directions respectively, obtain the distribution of the gray value k of the image, and calculate the corresponding gray sample variances S x and S y :

[0143] 。

[0144] 。

[0145] I32: Construct the target discrimination formula:

[0146] 。

[0147] I33: If Δ = 0, it is determined that the target light in the image is forward; if Δ ≠ 0, it is determined that the target light in the image is backward.

[0148] The above R1 also includes:

[0149] Evaluate the integrity of its light spot according to the variability E R combined with the light spot area A i For the light spots that meet the following conditions, they should be excluded:

[0150]

[0151] where is the variability threshold of the unoccluded light spot.

[0152] The decoding method further includes:

[0153] Calibrating the variability threshold to determine the initial value of the variability threshold:

[0154] Q1: Using the light targets arranged on the site as the calibration object, setting all the light targets to λ k wavelength by the remote host, where k ≤ b, b is the total number of wavelengths that each target lamp can emit, and all are lit.

[0155] Q2: Taking 1 image at the λ k wavelength, and obtaining the area A k,i of each light spot and the light spot variability E Rk,i ; i ≤ n, n is the total number of light spots in the image.

[0156] Q3: Keeping the corresponding relationship between E Rk,i and A k,i , arranging A k,i into a sequence {A k , n}, and obtaining the corresponding light spot variability sequence {E Rk,n}.

[0157] Q4: Searching for the minimum term E Rk,n in the sequence {E Rk,m , the corresponding term in the sequence {A k , n} is A k , m .

[0158] Q5: Searching for the maximum term E Rk,n in the sequence {E Rk,p , the corresponding term in the sequence {A k , n} is A k , p .

[0159] Q6: If A k , p in Q5 is the minimum in the sequence {A k , n}, then calculate the light spot variability threshold E k at the λ k,i wavelength for the light spot with area A R0(k，i) as follows:

[0160] .

[0161] Q7: If A k , p in Q5 is in the sequence {Ak , n If it is not the smallest in}, then this item is removed and Q5 is re-executed.

[0162] The above-mentioned R2 includes:

[0163] J1: Obtain the outer boundary and the target center coordinates of the target lamp images at each wavelength, and separately count the number of light spots in the images at each wavelength. The central dark area of the rear target lamp is not considered during the analysis.

[0164] J2: For the target lamp with b wavelengths, taking the wavelength image with the largest number of light spots as the reference, its wavelength is λ t , calculate the distance between the center of each light spot image at this wavelength and the center of the light spots in the images of other wavelengths respectively:

[0165] .

[0166] Among them, x k represents the abscissa of the k-th light spot center in the wavelength image with the largest number of light spots; y k represents the ordinate of the k-th light spot center in the wavelength image with the largest number of light spots; x λc,n represents the abscissa of the n-th light spot center in the wavelength image with the wavelength of λ c ; y λc,n represents the ordinate of the n-th light spot center in the wavelength image with the wavelength of λ c ; D(k, λ c , n) represents the center distance between the k-th light spot in the wavelength image with the largest number of light spots and the n-th light spot in the wavelength image with the wavelength of λ c .

[0167] J3: Among the numerous points with the wavelength of λ c , when a certain point can make the value of the previously calculated D(k, λ c , n) the smallest, then the distance between this point and the k-th point with the largest number of light spots and the wavelength of λ t is overlapped; using D(k, λ c , n), establish the corresponding relationship between each target point with the wavelength of λ t and the target points of other wavelengths according to the following formula:

[0168] .

[0169] Among them, P(λ t , k) represents the k-th point with the largest number of light spots and the wavelength of λ t ; P(λ c , q) represents the q-th point with the wavelength of λ c , and this point overlaps with P(λ t , k).

[0170] J4: Merge the center points of the light spots determined to be the same target lamp.

[0171] The J4 includes:

[0172] J41: Equivalently merge the center points of the light spots of different wavelengths of the same target lamp.

[0173] J42: Merge the areas of the light spots of different wavelengths of the same target lamp.

[0174] The J41 includes:

[0175] J411: Set the analysis accuracy: When calculating according to J412 to J418, when the search center gradually approaches the search center M point, the step size change will tend to 0; considering the discreteness of digital imaging, use Δ ≤ δ as the cut-off condition, where Δ is the step size and δ is the pixel size of the imaging sensor.

[0176] J412: Determine the initial search area: Determine the initial search range according to the coordinates of the center points P λ1 ~P λm (1 ≤ m ≤ b) of the light spots of different wavelengths of the same target lamp, and take min(x i ) ≤ x ≤ max(x i ), min(y i ) ≤ y ≤ max(y i ) (1 ≤ i ≤ m) area as the initial area.

[0177] J413: Determine the initial search center: Take the geometric center G of all the points in the area determined by J412 as the initial search center point M, and the coordinates are:

[0178] .

[0179] J414: Determine the initial search step size: Take half of the smaller value in the height and width directions of the area determined by J412 as the initial search step size Δ:

[0180] .

[0181] J415: Search area: With the optimal point M as the center and Δ as the step size, arrange search points in 8 orthogonal directions around it. For the first calculation, the M point is the G point; the coordinates of each of its search points are expressed as:

[0182] .

[0183] J416: Target calculation: For each search point arranged according to J415, calculate the distance between it and the center points of the light spots of m different wavelengths of the same target lamp respectively and sum them up:

[0184] 。

[0185] J417: According to J411, perform accuracy judgment. If it is satisfied, output the point position M, which is the optimal point; otherwise, adjust the search center and step size according to J418.

[0186] J418: For the point position M that does not meet the condition of J417, consider the following two situations:

[0187] Situation 1: If M coincides with the center point M in J415, it indicates that the initial search area is too large. At this time, update the step size Δ in J415 to Δ / 2 to further narrow the range; then recalculate according to the steps of J416 - J417.

[0188] Situation 2: If M does not coincide with the center point M in J415, at this time, the position of point M should be updated. After updating the point M in step J415 to this point, execute steps J416 - J417.

[0189] J419: Process the point P λ1 ~P λm according to the steps of J411 - J418 to complete the merging of the spot centers of different wavelengths of the same target lamp.

[0190] The said J42 includes:

[0191] J421: For the spots of different wavelengths of the same target lamp, calculate its merging center M(x M , y M ) according to J411 - J418, and then calculate the square S of the distance from the merging center M to the center of each wavelength spot respectively: i :

[0192] 。

[0193] J422: According to the minimum distance deviation D calculated by J416, calculate the distance deviation weight β between the spot centers of different wavelengths of the same target lamp and the merging center M:

[0194] 。

[0195] J423: According to the spot area A i of different wavelengths of the same target lamp, considering the influence of the distance deviation weight in J422, calculate the weighted average value of the superposition of the spot areas of each wavelength:

[0196] 。

[0197] J424: Process the spot areas corresponding to the points P λ1 ~P λm according to J421 - J423 to complete the merging of the spot areas of different wavelengths of the same target lamp.

[0198] The R3 includes:

[0199] R31: There are q pieces of data after merging the homologous target lights. The merging center of each homologous target image is (x Mi , y Mi ), and the merging area is A Ci (i ≤ q). The central dark area of the rear target light is not considered during analysis.

[0200] R32: Arrange A Ci in ascending order.

[0201] R33: Starting from A C1 , compare A Ci with A Ci+1 and calculate the relative deviation value δ i :

[0202] .

[0203] R34: If δ i > δ 0 , then classify A Ci+1 and A Ci as the same type of target, denoted as the r-th type.

[0204] Otherwise, classify A Ci+1 as the next type of target, denoted as the (r + 1)-th type.

[0205] Among them, δ 0 is the area resolution.

[0206] R35: If a total of t types of targets are classified according to the above R31 to R34, then discard the last group of grouped targets, that is, discard the t-th group, and retain the first t - 1 groups as the result of target clustering and grouping.

[0207] The decoding method further includes:

[0208] Calibrate the area resolution and determine the initial value of the area resolution:

[0209] P1: Use the optical targets arranged on the site as the calibration object, and set all the optical targets to the wavelength of λ k through the remote host, where k ≤ b, and b is the total number of wavelengths that each target light can emit, and all are lit.

[0210] P2: Take an image at the wavelength of λ k , and obtain the areas A k,i of the light spots in the image, where i ≤ n, and n is the total number of light spots in the image.

[0211] P3: Arrange A k,i in descending order to obtain the sequence {Ak , n , where n is the total number of light spots in the image, calculate the difference Δ between its adjacent terms k,j =A k , j+1 -A k , j ; j ≤ n, where n is the total number of light spots in the image.

[0212] P4: Take δ k , 0= min(Δ k,j ) and δ k , 0 > 0 value as the area resolution of the image at this wavelength.

[0213] P5: Calculate the δ k , 0 values of each light spot image at the b-type wavelengths respectively according to P1 - P4.

[0214] P6: Calculate the variability values of each light spot image at the b-type wavelengths respectively, and select the light spot with the smallest variability value as the calculation object.

[0215] P7: Take the light spot with the smallest variability value of each type of wavelength as the calculation object, and calculate the corresponding deviation weight β k .

[0216] P8: Weight-average the δ k , 0 values calculated in P5 according to the β k calculated in P7 to obtain the area resolution δ 0 :

[0217] .

[0218] The decoding method further includes:

[0219] Integrate the decoding results of the target lamp image set into an image label to obtain the first data.

[0220] The integration of the decoding results of the target lamp image set into an image label includes:

[0221] V1: Record the time stamp and camera number when the image is taken.

[0222] V2: Determine the viewing direction of the image and screen available target images, use coding 1 to represent forward and 0 to represent backward.

[0223] V3: Distinguish different target lamps according to the wavelength combination.

[0224] V4: Complete the coordinates of each target lamp image in the image plane, and represent each target lamp position with (x i, y i Record.

[0225] V5: Find all the target lights included in each target, and store the positions of the target lights in the same group as a set.

[0226] V6: Generate a coding sequence using the sets in V5 so that each target light image has a unique identifiable number.

[0227] The said V6 includes:

[0228] V61: If the determination in V2 is 1, for each set in V5, sort the x - coordinates of the target lights it contains in ascending order.

[0229] If the determination in V2 is 0, for each set in V5, sort the x - coordinates of the target lights it contains in descending order.

[0230] V62: Replace the coordinates corresponding to the target light sequence in V61 with the corresponding target light codes in V3 to obtain a new sequence, which is the entire target number.

[0231] The said decoding method further includes completing the autonomous identification of the target, and the steps are as follows:

[0232] N1: When triggering the camera to collect images, each wavelength of target light image will generate a single - channel image file in the camera and be transmitted to the remote host through in - field communication.

[0233] N2: The remote host calculates and stores the said image tags.

[0234] N3: Generate a visualized composite image.

[0235] N4: The remote host stores the "Unix timestamp" of the camera, the "camera number", and the "viewing direction", "band number", "target image code", "target image center" obtained in N2 as a data file DATA.

[0236] N5: Integrate each band image, the visualized composite image, and the data file DATA into 1 TIFF file.

[0237] The said N3 includes:

[0238] N31: Generate a blank image with the same resolution as each band image and the color level of each pixel in each channel being 0.

[0239] N32: Traverse each band image according to the following rules and assign values to the color levels of each pixel of the blank image generated in N31:

[0240] For any pixel (i, j) of each band image in the image plane coordinate system, if the gray level value of any one of its images is 255, then the gray level of the pixel (i, j) at the corresponding position in the blank image generated for N31 is assigned 255; otherwise, it is assigned 0.

[0241] A spatial multi-target image autonomous identification system, comprising:

[0242] The target as described above; the imaging device as described above; the coding device as described above.

[0243] The number of arranged targets is a first preset number, and they are arranged staggeredly along a preset baseline. The display directions of the lights of the same shape on all the targets are the same; the lights of adjacent targets on the same baseline are staggeredly distributed in the shooting direction of the imaging device.

[0244] The number of arranged imaging devices is a second preset number, which is used to collect the images of the lights on the targets.

[0245] The coding device is used to set the on-off combinations of the lights, encode the targets, and decode based on the images of the lights collected by the imaging device to complete autonomous identification.

[0246] A spatial multi-target image autonomous identification method, based on the above spatial multi-target image autonomous identification system, the method includes:

[0247] U1. Arrange the target cluster: Staggeredly arrange a first preset number of targets along a preset baseline. The display directions of the lights of the same shape on all the targets are the same; the lights of adjacent targets on the same baseline are staggeredly distributed in the shooting direction of the imaging device.

[0248] U2. Target coding: Set the on-off combinations of the lights on each target through the coding device to encode the targets.

[0249] U3. Collect the images of the lights and decode: Use the imaging device to collect the images of the lights on the targets to obtain a set of images of the lights; decode the set of images of the lights to obtain decoded data, and the decoded data includes the light direction, the image coordinates of the lights, and the number information of the targets to which the lights belong.

[0250] U4. Autonomous identification: Associate the decoded data with the set of images of the lights to complete autonomous identification.

[0251] In the U1, there are two or more preset baselines, and the preset baseline is a continuous space curve on any wall surface in the application scenario.

[0252] The present invention has the following beneficial effects:

[0253] The target for autonomous imaging identification of the present invention has a target rod for installing a target lamp adjustably installed on a target base. When applied to a large-scale cluster multi-target scenario, the angle of the target rod can be conveniently adjusted to prevent the images of the target lamps on adjacent targets from blocking each other in the camera; and all the target lamps can display lights on the front and back surfaces of the target rod, so as to cooperate with a camera cluster to achieve front-back bidirectional imaging in a large-scale cluster multi-target scenario.

[0254] The target lamp of the present invention has a light-shielding hole opened in the center of the front scattering cover or the rear scattering cover. One end of the reflection cone is provided with a light-shielding cylinder, and the light-shielding cylinder is sleeved in the light-shielding hole. The light-shielding cylinder is used to block part of the reflected light and shape the light spot. The purpose is to make the front and rear viewing directions of the target present different light spots to assist the camera in determining the viewing direction. There are counterbores for installing lamp beads on both the front and rear sides of the reflection cone. Each counterbore is slightly inclined, and the counterbores on the front and rear sides of the reflection cone are horizontally staggered and have the same number. Based on this structure, by reasonably setting the depth of the counterbores, the optical centers of all the lamp beads on the front and rear sides can be located on the same plane; compared with the existing planar direct illumination lamp bead layout, this structure realizes large-angle diffusion only through the light of a small number of light beads, and controls the aberration through a hyperboloid to ensure uniformity, thus avoiding increasing the number of lamp beads to make up for the lack of illumination uniformity and effectively reducing the heat generation of the device. The target lamp of the present invention can realize a large number of target information markings only through a small number of wavelength channels. Each target lamp contains three wavelength channel combinations, and n target lamp combinations can realize 3 n types of target markings, and the amount of information that can be marked grows exponentially with the number of target lamps, and the amount of information that can be expressed is huge.

[0255] The target fixing method of the present invention makes the entire target base contact the uneven surface of the installation structure through four leveling bolts. The ball head of the leveling bolt has good adaptability to the uneven surface, making the contact points firm and reliable. After installation, a firm connection relationship is formed between the entire target base and the installation structure; when the target base is fixed to a planar structure, the leveling bolts do not need to be installed. Therefore, this method is applicable to the surfaces of various structures and has more obvious advantages when installed on structures with uneven surfaces.

[0256] The equal-intensity parallax-free synchronous image splitting and multi-band imaging device of the present invention focuses the light beam through a lens and converges it on the focal plane behind the lens. First, a first-level beam splitter divides it into two beams. These two beams of light pass through the splitting channels and are split and reflected multiple times, and finally reach in front of multiple imaging components. At this time, the light intensity in front of each imaging component is equal and the images are parallax-free. Subsequently, the insensitive light beams of the imaging sensor are filtered by the filters of each imaging component to obtain parallax-free images in different wavelength channels. Since the light beams in front of each imaging component are all incident and split by the same lens, the parallax-free images can be ensured by accurately setting the position of the focusing lens. By accurately setting the position of the imaging sensor, the light beam can be accurately focused, so as to obtain clear and parallax-free images. Since the reflected and transmitted light intensities of the beam splitter are related to the incident angle, it is also difficult to ensure that the position and attitude of the beam splitter always remain fixed relative to the incident light based on the prior art. The equal-intensity parallax-free synchronous image splitting and multi-band imaging device of the present invention constructs a clever optical path, which can ensure that the incident angles of all beam splitters in the whole optical path are the same and the incident angles of all reflectors are the same for the light beam with any incident angle emitted from the focal plane, and an optical path compensation element is set to make the light intensity equal when the light beam reaches the imaging component.

[0257] The coding device and target coding control method for target coding of the present invention can use a remote host and a handheld terminal to code the target. Using the remote host can achieve large-scale coding operations, and using the handheld terminal can achieve fixed-point inspection and modification of the coding operations, which is convenient to operate.

[0258] The target coding method of the present invention, based on the display ability of the target lamp, establishes a mapping that can dynamically describe the correspondence between binary target coding and decimal target numbering through a conversion function, realizes the accurate conversion function between binary target coding and decimal target numbering, is convenient for coding modification and writing in the use stage, and brings great convenience to the use of users.

[0259] The decoding method of the target lamp image of the present invention determines the viewing direction by judging whether it is circular or annular on the premise of ensuring that the light spot is complete. The set process is as follows: First, define an index that can describe the difference between the afterimage and the complete target image, select a target image with the smallest degree of variation from the image for target type analysis, so as to complete the orientation of the viewing direction. During the decoding process, there is no need to separately deploy a target outside the field for calibration, and the calibration can be directly realized by using the targets arranged on-site; in addition, the variation threshold is adapted to the wavelength and the light spot area, taking into account the influence of the area difference of different bands and different imaging, and the analysis accuracy is higher; when merging the light spots from the same target lamp in the images of each wavelength, the problem of combination and matching of the target lamp images of each wavelength is solved. After the afterimage of the target lamp is removed, which light spots in the light spots of the images of each wavelength come from the same target lamp is screened out, so that the images of different wavelengths at the same position can be correctly combined together, and the target center after the combination of multi-band images is solved; the so-called merging is not simply to overlap these points, but to equivalently merge each point into the optimal point through the constructed algorithm. The characteristic of the merged optimal point is that the sum of the distances from the optimal point to each known point is the smallest, so as to ensure the accuracy of the merged data; when merging the light spot areas of different wavelengths of the same target lamp, it is not simply to add the areas of each image and then calculate the average value, but to consider the weight of the position distribution deviation and perform weighted averaging on each area, so that the superposition of images of multiple bands is all completed. This algorithm further improves the accuracy of the merged data; when classifying the target lamp images on the same target, an algorithm based on the aberration of the light spot area is adopted to classify the target lamp images with similar areas into the same target, improving the accuracy of the collection of target lamp images on the same target; the above method correlates and integrates the target lamp direction, the target lamp image coordinates, and the information of the target number to which it belongs without referring to the outside of a single image, realizing the autonomous identification of the target image; through a TIFF file, that is, integrating the data information of each band image and its joint processing, this data processing method can effectively solve the problems of large quantity of image acquisition by each camera and difficult storage and classification, and because it contains all the information starting from the original image, its traceability is strong and it is convenient for later verification and proofreading.

[0260] The spatial multi-target image autonomous identification system and the spatial multi-target image autonomous identification method of the present invention include the above-mentioned targets, imaging devices, encoding devices and their related control methods, and therefore have corresponding technical effects, which will not be elaborated here.

[0261] In addition to the purposes, features and advantages described above, the present invention has other purposes, features and advantages. The present invention will be further described in detail below with reference to the drawings. Description of the Drawings

[0262] The accompanying drawings, which form a part of this application, are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention.

[0263] Figure 1 It is a schematic diagram of a large-scale cluster camera and a multi-target scenario.

[0264] Figure 2 It is a target image map identified by a selected camera in a large-scale cluster camera and multi-target scenario.

[0265] Figure 3 It is a target image map identified after the dynamic displacement of a selected camera in a large-scale cluster camera and multi-target scenario.

[0266] Figure 4 It is a schematic diagram of the shooting state of a multi-lens multi-filter camera.

[0267] Figure 5 It is a schematic diagram of the parallax between multi-band images obtained by a multi-lens multi-filter camera.

[0268] Figure 6 It is a schematic diagram of the shooting state of a single-lens multi-filter camera.

[0269] Figure 7 It is a schematic diagram of the parallax between multi-band images obtained by a single-lens multi-filter camera.

[0270] Figure 8 It is a schematic diagram of the parallax between multi-band images obtained by an existing optical image splitting device.

[0271] Figure 9 It is a schematic diagram of the three-dimensional structure of the target from the first perspective in the target embodiment of the present invention.

[0272] Figure 10 It is a schematic diagram of the three-dimensional structure of the target from the second perspective in the target embodiment of the present invention.

[0273] Figure 11 It is a schematic diagram of the disassembled three-dimensional structure of the target in the target embodiment of the present invention.

[0274] Figure 12 It is a schematic diagram of the three-dimensional structure of the target rod in the target embodiment of the present invention.

[0275] Figure 13 It is a schematic diagram of the disassembled three-dimensional structure of the target base from the first perspective in the target embodiment of the present invention.

[0276] Figure 14 It is a schematic diagram of the disassembled three-dimensional structure of the target base from the second perspective in the target embodiment of the present invention.

[0277] Figure 15It is a three-dimensional structural schematic diagram of the target lamp and a corresponding radial sectional three-dimensional schematic diagram in the target lamp embodiment of the present invention.

[0278] Figure 16 It is a disassembled three-dimensional structural schematic diagram of the target lamp in the target lamp embodiment of the present invention.

[0279] Figure 17 It is a three-dimensional structural schematic diagram of the reflection cone in the target lamp embodiment of the present invention.

[0280] Figure 18 It is a layout schematic diagram of the lamp beads on the reflection cone in the target lamp embodiment of the present invention.

[0281] Figure 19 It is a comparison diagram of the illumination effects between the lamp beads in the target lamp embodiment of the present invention and traditional lamp beads.

[0282] Figure 20 It is a flow schematic diagram of the target fixing method embodiment of the present invention.

[0283] Figure 21 It is a structural schematic diagram of the target fixed to the installation structure.

[0284] Figure 22 It is a sectional three-dimensional structural schematic diagram of the imaging device embodiment of the present invention.

[0285] Figure 23 It is a sectional structural schematic diagram of the imaging device embodiment of the present invention.

[0286] Figure 24 It is a schematic diagram of the application scenario simulation of the imaging device embodiment of the present invention.

[0287] Figure 25 It is a schematic diagram of the multi-band split image taken by the imaging device embodiment of the present invention.

[0288] Figure 26 It is a control system diagram of the coding device embodiment of the present invention.

[0289] Figure 27 It is a network topology structure schematic diagram of the coding device embodiment of the present invention.

[0290] Figure 28 It is a schematic diagram of the illumination distribution when different numbers of lamp beads are damaged.

[0291] Figure 29 It is a schematic diagram of the coding method embodiment of the present invention.

[0292] Figure 30 It is a flow schematic diagram of converting numbers to codes in the coding method embodiment of the present invention.

[0293] Figure 31It is a schematic diagram of the process of encoding conversion to numbers in the encoding method embodiment of the present invention.

[0294] Figure 32 It is a schematic diagram of the afterimage of the target lamp in the decoding method embodiment of the target lamp image of the present invention.

[0295] Figure 33 It is a schematic diagram of the gray-scale distribution of the center of the target lamp in the decoding method embodiment of the target lamp image of the present invention.

[0296] Figure 34 It is a schematic diagram of the principle of multi-band image merging in the decoding method embodiment of the target lamp image of the present invention.

[0297] Figure 35 It is a schematic diagram of the traversal of the center distance of the multi-band target lamp image in the decoding method embodiment of the target lamp image of the present invention.

[0298] Figure 36 It is a schematic diagram of the process of merging the centers of the multi-band target lamp images in the decoding method embodiment of the target lamp image of the present invention.

[0299] Figure 37 It is a schematic diagram of the clustering of the imaging of the target lamp of the same target in the decoding method embodiment of the target lamp image of the present invention.

[0300] Figure 38 It is a schematic diagram of the generation of image labels in the decoding method embodiment of the target lamp image of the present invention.

[0301] Figure 39 It is a schematic diagram of the generation of label image files in the decoding method embodiment of the target lamp image of the present invention.

[0302] Each label in the drawings represents:

[0303] 100, target pole; 200, target lamp; 201, backscattering cover; 202, back hyperboloid reflective cavity; 203, reflective cone; 204, lamp beads; 205, front hyperboloid reflective cavity; 206, front scattering cover; 207, light shielding tube; 300, target base; 301, arc guide groove; 302, locking screw; 303, chassis; 3031, fixing fastener; 3032, leveling bolt; 3033, mounting through hole; 3034, leveling screw hole; 400, lens; 500, rack; 501, light splitting channel; 5011, primary beam splitter; 5012, secondary beam splitter; 5013, secondary beam splitter; 5014, residual light absorption cavity; 502, imaging component; 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

[0304] The embodiments of the present invention are described in detail below with reference to the accompanying drawings, but the present invention can be implemented in many different ways as defined and covered by the claims.

[0305] Target Examples:

[0306] like Figures 9 to 12 As 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 by 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 applied to 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 in the images formed 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 cooperation with a camera group in a large-scale cluster multi-target scene.

[0307] As Figure 13 , Figure 14 shown, in this embodiment, the target base 300 is provided with a chassis 303, which provides support for the rotation of the target rod and the connection with the structure, and is integrally formed with lightweight aluminum alloy; a set of mounting and fixing components are respectively provided at both ends of the chassis 303. The mounting and fixing components include a fixed fastener 3031 and a pair of leveling bolts 3032. The target base 300 is provided with a mounting through hole 3033 for mounting the fixed fastener 3031, and leveling screw holes 3034 for mounting a pair of leveling bolts 3032 are respectively provided on both sides of the mounting through hole 3033. The bottom of the leveling bolt 3032 is spherical, and the fixed 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 bolt 3032 abuts against the surface of the installation 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 chassis 303 and the installation structure can be controlled. After the chassis 303 is basically in the required installation state, the target base 300 is fixed to the surface of the installation structure by installing the fixed fastener 3031.

[0308] As Figure 13 , Figure 14 shown, in this embodiment, an arc guide groove 301 is provided on the chassis 303. The target rod 100 is hinged to the target base 300, and the hinge point is located at the center of the arc of the arc guide groove 301. 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 wing nut, and the rotation of the target rod 100 can be controlled by tightening or loosening the wing nut.

[0309] In this embodiment, three target lights 200 are provided. The three target lights 200 are evenly spaced along the target rod 100. The three target lights 200 can all provide light sources of multiple bands, and the lights shown on the front and back sides of the three target lights 200 are the same in the same direction and different in the opposite direction.

[0310] Embodiment of the target light:

[0311] As Figures 15 to 18As shown in the figure, the target lamp 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 docked with the front hyperbolic reflection cavity 205 and clamps and fixes the reflection cone 203 inside. There are multiple lamp beads 204, which are scattered and arranged on the front and rear 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 and equalizes the light reflected backward, and the front scattering cover 206 scatters and equalizes the light reflected forward. Both the rear hyperbolic reflection cavity 202 and the front hyperbolic reflection cavity 205 are integrally formed by lightweight aluminum alloy, which is convenient for heat dissipation of the light source. The rear hyperbolic reflection cavity 202 shapes the light path according to a certain rule and reflects it backward to the target, and the front hyperbolic reflection cavity 205 shapes the light path according to a certain rule and reflects it forward to the target. The reflection cone 203 turns the light emitted by the lamp beads 204 in the reflection cavity in the front and rear directions respectively. The inner side surfaces of the rear scattering cover 201 and the front scattering cover 206 are spherical, and the outer side surfaces are flat. Both the inner and outer side surfaces are treated with a frosted process. A card slot for installing the rear scattering cover 201 is provided on the hyperbolic reflection cavity 202, and the card 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 surfaces formed by rotating a hyperbola around the central axis of the target lamp, and their surfaces are coated with a metal plating mirror surface.

[0312] In any embodiment, a light-shielding hole is opened in the center of the front scattering cover 206 or the rear scattering cover 201. One end of the reflection cone 203 is provided with a light-shielding cylinder 207, and the light-shielding cylinder 207 is sleeved in the light-shielding hole. In this embodiment, it is selected to open the light-shielding hole on the rear scattering cover 201. The light-shielding cylinder 207 on the reflection cone 203 has an inverted frustum structure, is located at the top on the rear side of the reflection cone 203, and cooperates with the light-shielding hole on the rear scattering cover 201. The light-shielding cylinder 207 is used to block part of the reflected light and shape the light spot, aiming to make the front and rear viewing directions of the target present different light spots to assist the camera in determining the viewing direction.

[0313] In any embodiment, the lamp beads 204 include LED lamp beads of multiple different wavelengths. Multiple different wavelengths of LED lamp beads are provided on both the front and rear sides of the reflection cone 203. The multiple different wavelengths of LED lamp beads on the same side of the reflection cone 203 are arranged staggeredly. 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 requirements of sensor sensitivity differences.

[0314] In this embodiment, the lamp beads 204 on the front and rear sides of the reflection cone 203 are arranged staggeredly. The optical centers of the lamp beads 204 on the front and rear sides of the reflection cone 203 are located on the same plane. The rear hyperbolic reflection cavity 202, the front hyperbolic reflection cavity 205, and the reflection cone 203 are coaxially arranged. On both the front and rear sides of the reflection cone 203, there are counterbores for installing the lamp beads 204. Each counterbore is slightly inclined, and the counterbores on the front and rear sides of the reflection cone 203 are horizontally staggered and have equal numbers. Based on this structure, by reasonably setting the depth of the counterbores, the optical centers of all the lamp beads 204 on the front and rear sides can be located on the same plane. Compared with the existing planar direct-emitting lamp bead layout, this structure realizes large-angle diffusion only through the light rays of a small number of light beads, and controls the aberration through the hyperbolic surface to ensure uniformity, thereby avoiding increasing the number of lamp beads to make up for the lack of illumination uniformity and effectively reducing the heat generation of the device. A tray is provided on the outer periphery of the reflection cone 203, and this tray is used to cooperate with the rear hyperbolic reflection cavity 202 and the front hyperbolic reflection cavity 205 to limit and fix the entire reflection cone 203.

[0315] Based on the target lamp structure of the present invention, through the cooperation of the rear scattering cover 201 and the rear hyperbolic reflection cavity 202, and the cooperation of the front scattering cover 206 and the front hyperbolic reflection cavity 205, the point light sources formed by each lamp bead 204 can be made equivalent to a single point light source, so that the optical paths of the light rays emitted by each lamp bead 204 reaching the scattering cover after being reflected by the inner surface of the reflection cavity are equal, enabling it to have good spatial homology and uniformity. As Figure 19 shown, through simulation by an optical simulation software, its illumination effect is more uniform than that of traditional lamp beads.

[0316] Based on the red, green, and blue LED lamp beads with different wavelengths in this embodiment, by controlling the on-off combinations of the lamp beads 204 with different wavelengths, a single target lamp can form seven different combined output results, as shown in Table 1 specifically. Table 1 is: the seven different combined output results formed by a single target lamp.

[0317]

[0318] The combined output results on the front and rear sides of the target lamp have the same color but different shapes. Due to the cooperation of the light-shielding holes on the rear scattering cover 201 and the light-shielding cylinder 207 on the reflection cone 203, part of the reflected light is blocked, making the light spot formed on the rear side of the target lamp an annular light spot, while there is no blockage on the front side of the target lamp, so it is a circular light spot, thereby enabling the rapid determination of the camera's viewing direction. The target lamp in this embodiment can realize a large number of target information markings through only a small number of wavelength channels. Each target lamp contains 7 wavelength channel combinations, and n target lamp combinations can achieve 7 n types of target markings. The amount of information that can be marked and the number of target lamps increase exponentially, and the amount of information that can be expressed is huge.

[0319] In practical applications, after all the targets are pre-coded by the coding device of the present invention, the on / off states of the target lights on each target are uniformly set and allocated. The specific coding device, coding method, etc. are as described in the following embodiments.

[0320] Embodiment of the target fixing method:

[0321] This embodiment is a fixing method for fixing the target for image autonomous identification on an uneven surface based on the above-mentioned target embodiment. The flowchart of this method is as Figure 20 shown, and the installation structure is as Figure 21 shown. Specifically, it includes the following steps:

[0322] S1: Pre-installation. Pass the fixing fastener 3031 through the installation through-hole 3033 on the target base 300 and drive it into the uneven surface, and make the fixing fastener 3031 not fully compress the target base 300. Since the bottom surface of the target base 300 is a plane and cannot be closely attached to the surface of the installation structure, there are unstable contact points between the bottom surface of the target base 300 and the surface of the installation structure. At this time, the target base 300 is still prone to displacement under external disturbances, and then enter the next step of installation.

[0323] S2: Leveling. Screw the leveling bolts 3032 into the respective leveling screw holes 3034 so that the spherical parts at the bottoms of the leveling bolts 3032 are in contact with and abutted against the uneven surface, and adjust the screwing depth of each leveling bolt 3032 to make the attitude of the target base 300 tend to the target installation attitude.

[0324] S3: Fixing. When the target base 300 reaches the target installation attitude, tighten the fixing fastener 3031 to completely fix the target base 300 and complete the fixed installation of the target on the uneven surface.

[0325] Adopting the method of this embodiment, the entire target base 300 is in contact with the uneven surface of the installation structure through 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 installation, a stable connection relationship is formed between the entire target base 300 and the installation structure; when the target base 300 is fixed to a flat structure, the leveling bolts 3032 do not need to be installed. Therefore, this method can be applied to the surfaces of various structures and has more obvious advantages when installed on the structures with uneven surfaces.

[0326] Embodiment of the imaging device:

[0327] As Figures 22 to 23As shown in the figure, the equal-intensity parallax-free synchronous split imaging 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, and the lens 400 is used to focus the light beam of the shooting object on the focal plane of the lens 400. A splitting channel 501 and multiple imaging components 502 are provided in the frame 500. The splitting channel 501 is a cavity for the main optical path transmission, and light-absorbing coatings are applied to its surfaces. The splitting channel 501 forms equal-intensity light by multi-stage splitting and reflection of the light incident from the lens 400 and transmits it to each imaging component 502. The imaging component 502 includes a focusing lens 5021, an imaging sensor 5023, and a band filter 5022 for allowing a specific wavelength light beam to pass through. The focusing lens 5021 is arranged in front of the filter 5022 to refocus each light beam onto the imaging sensor 5023 for imaging. Each focusing lens 5021 is parallel to each other and perpendicular to the optical axis of the lens 400.

[0328] In this embodiment, there are four groups of imaging components 502, and the splitting channel 501 is provided with one first-stage beam splitter 5011 and two second-stage beam splitters 5012. After the incident light is dispersed into two light beams by the first-stage beam splitter 5011, the two light beams are respectively dispersed into two light beams by one second-stage beam splitter 5012, forming a total of four light beams that are respectively transmitted to the four groups of imaging components 502. In this embodiment, convex lenses that have undergone chromatic aberration correction are used as the focusing lenses 5021, and the optical properties and sizes of each focusing lens 5021 are the same; chromatic aberration is a type of aberration. When light beams of different wavelengths are focused by an optical system, their imaging positions are not on the same plane. In this embodiment, four wavelengths of light need to be focused, and the focusing lens 5021 will cause the light beams of different wavelengths to not be focused on the same plane position due to chromatic aberration. Therefore, chromatic aberration correction is required to reduce this difference; in this embodiment, there are a total of four filters 5022, all of which are perpendicular to the optical axis of the lens 400, and the allowable wavelength ranges of each filter 5022 do not overlap. The imaging sensor 5023 is an electrical component used to receive each refocused light beam and image. In this embodiment, there are a total of four imaging sensors 5023, and they are only sensitive to the allowable wavelengths of the corresponding filters 5022, used to output images of four different bands; each imaging sensor 5023 is perpendicular to the optical axis of the lens 400 and is provided with a device for precise fine-tuning along its axis to correct the chromatic aberration of the light beams of different wavelengths focused here; this device separates and images light sources of various bands. Therefore, in addition to correcting the chromatic aberration of the lens, this embodiment also achieves precise imaging through the method of fine-tuning the imaging plane.

[0329] In this embodiment, the incident light is dispersed by a primary beam splitter 5011 into a first light beam L1 and a second light beam L2. The first light beam L1 is reflected by a first reflector M1 and then directed towards a secondary beam splitter 5012. After being dispersed by the secondary beam splitter 5012 into two light beams, they are respectively reflected by a second reflector M2 and a third reflector M3 and then enter the corresponding imaging assembly 502. The second light beam L2 directly strikes another secondary beam splitter 5012. After being dispersed by the secondary beam splitter 5012 into two light beams, one of them is successively reflected by a fourth reflector M4 and a fifth reflector M5 and then enters the corresponding imaging assembly 502, and the other is successively reflected by a sixth reflector M6 and a seventh reflector M7 and then enters the corresponding imaging assembly 502.

[0330] In this embodiment, a secondary beam splitter 5013 and a stray light absorption cavity 5014 are provided between the secondary beam splitter 5012 and the imaging assembly 502. The stray light absorption cavity 5014 is a cavity used to absorb the excess transmitted or reflected light generated during the optical path compensation process. The inner surface of the stray light absorption cavity 5014 is coated with an absorbent coating to prevent the stray light from interfering with other components when the beam splitters are close to other optical elements.

[0331] In this embodiment, the primary beam splitter 5011, the secondary beam splitter 5012, and the secondary beam splitter 5013 are arranged in parallel. All the beam splitters are used to decompose a beam of incident light into a reflected light beam and a transmitted light beam, and the optical properties and thicknesses of the beam splitters in this embodiment are exactly the same. The first reflector M1, the fourth reflector M4, and the sixth reflector M6 are arranged in parallel and are parallel to the beam splitters; the second reflector M2, the third reflector M3, the fifth reflector M5, and the seventh reflector M7 are arranged in parallel and are perpendicular to the beam splitters. All the reflectors have exactly the same optical properties.

[0332] In this embodiment, a plurality of heat insulation cavities 503 are provided inside and in the middle of the outer shell surface of the frame 500. The plurality of heat insulation cavities 503 are separated by a plurality of stiffening ribs. The stiffening ribs, as subordinate components of the frame 500, can improve the stiffness of the frame 500, ensure that the frame 500 does not deform. The heat insulation cavities 503 can achieve a heat insulation effect, reduce the deformation of the frame 500 caused by temperature changes, prevent the position change of the optical elements, and ensure the stability of the optical path.

[0333] 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 it on the focal plane behind the lens. First, it uses a first beam splitter 5011 to split it into two beams. These two beams of light pass through the beam splitting channels 501 and are split and reflected multiple times, and finally reach in front of multiple imaging components 502. At this time, the light intensity in front of each imaging component 502 is equal and the images are parallax-free. Subsequently, the filters 5022 of each imaging component 502 filter out the insensitive light beams of the imaging sensors 5023 to obtain parallax-free images in different wavelength channels. Since the light beams in front of each imaging component 502 are all incident and split by the same lens 400, by precisely setting the position of the focusing lens 5021, it is possible to ensure that the images are parallax-free. By precisely setting the position of the imaging sensor 5023, it is possible to ensure that the light beam is accurately focused, thereby obtaining clear and parallax-free images (such as Figure 24 , Figure 25 shown). Since the reflected and transmitted light intensities of the beam splitter are related to the incident angle, it is also very difficult to ensure that the position and attitude of the beam splitter always remain fixed relative to the incident light based on the existing technology. The equal-intensity parallax-free synchronous image splitting and multi-band imaging device of the present invention constructs a clever optical path, which can ensure that the incident angles of all beam splitters and all reflectors in the entire optical path are the same for light beams with any incident angle emitted from the focal plane, and an optical path compensation element is set to make the light intensity equal when the light beam reaches the imaging component 502.

[0334] Embodiment of the encoding device:

[0335] As Figure 26 , Figure 27 shown, the encoding device for target encoding in this embodiment includes a remote host 600, more than one handheld terminal 601, and more than one target encoding module 602. The handheld terminal 601 is used to communicate with the remote host 600 for database synchronization, and is also used to communicate with the target encoding module 602 and perform addition, deletion, query, and modification of 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 encoding and status information of the target to the handheld terminal 601.

[0336] 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 WIFI unit 6014; the first radio frequency unit 6011 includes an RFID tag for communicating with the target coding module 602; the first storage unit 6012 is used to store the data of the handheld terminal 601; the first computing unit 6013 is used for calculating the coding information of the handheld terminal 601; the WIFI unit 6014 is used for communicating with the remote host 600. The main functions of the handheld terminal 601 are the on-site installation and commissioning of the target, the initialization and modification of the coding, and the synchronization with the database of the remote host 600 through the WIFI unit 6014 to ensure the accuracy and timeliness of the database of the handheld terminal 601; the communication with the target coding module 602 through the first radio frequency unit 6011 to realize the addition, deletion, query, and modification of the target coding information on-site.

[0337] In this embodiment, the target coding module 602 is separately integrated into the target and includes a second radio frequency unit 6021, a second storage unit 6022, a second computing unit 6023, and a NET 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 the data of the target coding module 602; the second computing unit 6023 is used for calculating the coding information of the target coding module 602; the NET unit 6024 is used for communicating with the remote host 600. Communication between the handheld terminal 601 and the target coding module 602 is carried out through the first radio frequency unit 6011 and the second radio frequency unit 6021, and this communication is realized through the in-site local area network. The real-time coding information of the target is synchronized to the remote host 600 through the NET unit 6024; communication is carried out with the first radio frequency unit 6011 of the handheld terminal 601 through the second radio frequency unit 6021 to receive the deletion and modification instructions of the coding from the handheld terminal 601 and upload the coding and status information of the target to it.

[0338] In this embodiment, the network relationships between the remote host 600 and one or more target coding modules 602 include: the remote host 600 can perform unidirectional read and write operations on the data of one or more target coding modules 602. The network relationships between the remote host 600 and one or more handheld terminals 601 include: the remote host 600 can perform read and write operations on the data of one or more handheld terminals 601; each handheld terminal 601 can perform read operations on the data of the remote host 600. The remote host 600 can publish information to the outside through the 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 coding module 602 includes: the handheld terminal 601 can perform data reading operations on the target coding module 602. Before the handheld terminal 601 performs data reading operations on the target coding module 602, the handheld terminal 601 needs to communicate with the remote host 600 for database synchronization first.

[0339] Embodiment of the target coding control method:

[0340] The target coding control method of this embodiment is based on the foregoing coding device and target lamp embodiments. This method includes:

[0341] T1. Communicate with the remote host 600 through the handheld terminal 601 to synchronize the database, communicate with the target coding module 602, and perform addition, deletion, query, and modification of the target coding information.

[0342] T2. Communicate with the handheld terminal 601 through the target coding module 602 to receive the addition, deletion, query, and modification instructions from the handheld terminal 601; send the coding and status information of the target to the handheld terminal 601 through the target coding module 602.

[0343] In this embodiment, T1 and T2 are mainly used for querying, initializing, and modifying the target coding on-site. Among them, T1 includes:

[0344] A1. For the target with initialized or modified coding, use the handheld terminal 601 to synchronize the coding library with the remote host 600 through the WIFI unit 6014 and update the coding library in the first storage unit 6012.

[0345] A2. Input the proposed target number through the handheld terminal 601 and perform coding through the first calculation unit 6013.

[0346] A3. The first calculation unit 6013 reads the data in the first storage unit 6012 and compares the input and encoded target number with it. If it duplicates the existing coding, it returns for re-input; if not, it writes the input and encoded target number into the first storage unit 6012.

[0347] 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.

[0348] A5. When the RFID tag of the second radio frequency unit 6021 receives the coding write instruction, trigger the read and write instruction of the second calculation unit 6023, and the second storage unit 6022 receives and writes the new coding.

[0349] A6. The new coding written by the second storage unit 6022 is synchronized to the remote host 600 through the NET unit 6024, and its coding library is updated.

[0350] A7. The second calculation unit 6023 outputs the new coding of the second storage unit 6022 to the corresponding lamp beads 204 of each target lamp to control the on / off combination of the lamp beads 204.

[0351] In this embodiment, step T2 includes:

[0352] 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.

[0353] B2: When the RFID tag of the second RF unit 6021 on the target receives a query instruction, it triggers the read / write function of the second calculation unit 6023 to read the latest target coding information and working status information in the second storage unit 6022.

[0354] B3: The RFID tag of the second RF unit 6021 on the target feeds back the latest target coding information and working status information to the RFID tag of the first RF unit 6011 on the handheld terminal 601, and then decodes it through the first calculation unit 6013 and displays it on the handheld terminal 601.

[0355] In this embodiment, the coding information of the target can also be read, written, collected, and corrected by the remote host 600, and the on / off combination of the target lights can be adjusted. This part of the method is mainly used for remote and large-scale collection and correction of target coding, and specifically includes:

[0356] C1: The remote host 600 sends a coding information collection instruction to each target.

[0357] C2: Trigger the second calculation unit 6023 of each target to execute the coding information read / write instruction, and send the coding information stored in the second storage unit 6022 to the remote host 600 through the NET unit 6024.

[0358] C3: The remote host 600 verifies the collected target coding information. If there is a duplicate code situation, the duplicate code is corrected, a code correction instruction is sent to the corresponding target, and the corrected coding information is sent to the corresponding target.

[0359] C4: The second calculation unit 6023 of the corresponding target receives the code correction instruction sent by the remote host 600 and the corrected coding information, and writes the corrected coding information into the second storage unit 6022 for refreshing.

[0360] C5: The second calculation unit 6023 controls the corresponding lamp beads 204 of the target light according to the corrected coding information to adjust its on / off combination.

[0361] The target expresses information through the combination of the lighting and extinguishing of the lamp beads 204. If a lamp bead 204 fails and shows a lighting and extinguishing situation inconsistent with the coding, for example, if a lamp bead 204 of a certain wavelength is damaged, it will cause the wavelength combination appearing in the image to be inconsistent with the actual coding, resulting in an error in the system's encoding and decoding of the image; and when the system power-off occurs, after the power is restored, the target lamp needs to be re-lit, and it needs to be restored to the preset state. In this embodiment, there are also two control modes: status monitoring and power-off recovery. Status monitoring: When a fault occurs in the target lamp, the coding device is used to monitor the status of the target; Power-off recovery: When the system experiences a power-off and then power-on situation and the target lamp returns to the default state, the coding device is used to restart the target lamp to restore the preset state of the target.

[0362] In this embodiment, the specific steps of status monitoring include:

[0363] D1: Optical simulation is adopted to calculate the most unfavorable light distribution with different numbers of damaged lamp beads 204. Considering the light uniformity and the minimum light resolution of the camera comprehensively, the allowable number of damaged lamp beads 204 is determined; Figure 28 It shows the illuminance distribution with different numbers of damaged lamp beads 204 when the number of single-wavelength lamp beads 204 is four. As can be seen from Figure (c), when the number of damaged lamp beads reaches two (the figure shows the most unfavorable working condition with two adjacent damaged lamp beads, and at this time the light is seriously eccentric), a significant low-illuminance distribution area appears in the light distribution. Therefore, when the number of single-wavelength lamp beads 204 is four, one damaged lamp bead 204 is allowed. This method takes into account the engineering reality and economy, allows a certain number of damaged lamp beads 204 of the same wavelength in each target lamp, and the allowable number of damaged lamp beads is determined according to the above optical simulation method.

[0364] D2: The second calculation unit 6023 collects the lighting and extinguishing states of each lamp bead 204 at a preset frequency and generates corresponding monitoring codes; for the lamp beads 204 of a certain wavelength in the same target lamp, when the number of damaged lamp beads 204 collected is the allowable number in D1, the lamp bead 204 is still recorded as fully lit, otherwise it is recorded as fully extinguished; when the lamp bead 204 is marked as the fully extinguished state, 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 the lighting combination and one is monitored to be damaged, the system still considers that all four are fully lit at this time, 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 monitored to be damaged, the system considers that all four lamp beads 204 are fully extinguished at this time, and its corresponding code is recorded as 0, and the monitoring code is written into the second storage unit 6022.

[0365] D3. Compare the monitoring code generated in D2 with the original code in the second storage unit 6022. If the comparison result is consistent, it is determined that the wavelength lamp bead 204 is normal and the current state is maintained; if the comparison result is inconsistent, it is determined that the wavelength lamp bead 204 is damaged, and the second calculation unit 6023 sends an instruction to abort illumination, and all the target lamps of the target are turned off.

[0366] 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 alarms for maintenance.

[0367] D5: The remote host 600 publishes maintenance information to the off-site terminal through the mobile network.

[0368] In this embodiment, for power failure recovery, it should be noted that according to the foregoing control logic, there is a possibility that the handheld terminal 601 and the remote host 600 read and write the target data simultaneously when encoding the target (the read and write frequency of the remote host 600 is much higher than that of the handheld terminal 601). To ensure the accuracy and reliability of the encoded information stored in the target, in this embodiment, the RFID tag of the target is only used for temporary storage during communication and encoding with the handheld terminal 601, and the RFID tag of the target is not used as a permanent storage medium. The data processed by the second calculation unit 6023 of the target is saved using a separate second storage unit 6022. Therefore, when a power failure occurs in the field, after the power is restored, the preset state of the target is restored by reading the second storage unit 6022 instead of the target RFID tag.

[0369] The specific steps for power failure recovery in this embodiment include:

[0370] E1: Before the power failure, the second storage unit 6022 has stored the target encoding information.

[0371] E2: The second calculation unit 6023 monitors the power restoration information and performs a self-check on the target lamps according to D1 to D5.

[0372] E3: For the target lamps that pass the self-check, the second calculation unit 6023 outputs the target encoding of the second storage unit 6022 to the corresponding lamp beads 204 of the target lamps to restore the on / off combination of the lamp beads 204.

[0373] Embodiment of the target encoding method:

[0374] A target encoding method, based on the above-mentioned target and target lamp embodiments, assuming that there are a target lamps on a single target, and each target lamp has b types of wavelength lamp beads 204. Excluding the target with all target lamps turned off, each target lamp has a total of 2 b -1 combinations; and each target has a target lamps, so for a target with a target lamps and each target lamp having b wavelengths, the number of target numbers that can be expressed is: (2b -1) a In this embodiment, a = 3, b = 3, and each target lamp has 2 3 -1 = 7 combinations, and the number of target numbers that can be expressed is: (2 3 -1) 3 = 343 kinds.

[0375] As Figures 29 to 31 shown, the target coding method includes the following steps:

[0376] F1: Number of bits for encoding a single target: Each target lamp has b kinds of wavelength lamp beads 204. Each kind of wavelength lamp bead 204 needs to occupy 1 bit of encoding to represent 2 states of on or off, where 1 represents on and 0 represents off. Therefore, each target lamp requires b bits of encoding, and a target lamps of a single target require a·b bits of encoding; in this embodiment, each target lamp has 3-bit encoding, and 3 target lamps of a single target are 3·3 = 9-bit encoding.

[0377] F2: Encoding of a single target lamp: For any one target lamp i on the target, connect the on-off status codes of b kinds of wavelengths in sequence to form the target lamp encoding to represent the identity of the current target lamp. Each target lamp encoding has b bits; as Figure 29 the target lamp 2 in 1 , its on-off states of 3 wavelengths λ 2 , λ b are 1, 1, 1 respectively, and connecting them forms the encoding of a single target lamp as 111.

[0378] F3: Encoding of a single target: Connect the encodings of a single target lamp from the first target lamp to the bth target lamp in sequence from left to right to obtain the encoding of the entire target; in this embodiment, this sequence is globally unified. Even if the encoding is reversed in the backward encoding, it is read in this sequence during the decoding of the encoding image, and connect them in sequence to obtain the encoding of the entire target. As Figure 29 in the forward target in

[0379]

[0380] Since the target encoding is binary, it is suitable for computer processing, but it is not convenient for encoding modification and writing during use. Therefore, we need to design a transcoding method to meet the following conditions:

[0381] ⅰ The numbers input by the user are consecutive decimals;

[0382] ⅱ The numbers input by the user terminal are in one-to-one correspondence with the codes described in F3. It should be noted here that not all decimal numbers can be mapped to the codes described in F3. When converting a decimal number to a binary code, there may be consecutive 0s. For example, when 400 is converted to a binary code, it is 100 000 000. According to the coding rule of F3, in this case, only the first target light is lit and the remaining two are extinguished. However, this situation is not allowed. Therefore, when establishing the correspondence, such a situation should be excluded.

[0383] ⅲ The correspondence in ⅱ above should not be a search of a pre-built form, but rather a search for a mapping that can dynamically describe the correspondence, such as a conversion function. Because the pre-built form is based on an exhaustive approach, on the one hand, its flexibility is poor, it requires a large amount of storage, and it is not easy to maintain; on the other hand, it cannot essentially reflect the correspondence and cannot solve the problem of non-consecutive decimal encodings mentioned in ⅱ, which will cause the user terminal to avoid some numbers, bringing great inconvenience to the user.

[0384] To address the above issues, step F4 of this embodiment includes two parts. One is to convert the target number into a target code, and the other is to convert the target code into a target number.

[0385] In this embodiment, converting the target number into a target code includes:

[0386] G1: Assume that the user inputs the decimal number k of the target to be encoded through the handheld terminal 601 or the remote host 600 10 , and compare it with the existing codes according to steps A1 to A7. If it is repeated with the existing codes, return a prompt to re-enter; otherwise, proceed to G2.

[0387] G2: Assume that each target has a target lights, and each target light has b wavelengths. The first calculation unit 6013 or the remote host 600 converts the decimal number K 10 into a binary code K b . In this embodiment, each target light has 3 wavelengths, so K 2b is converted into an octal code K 10 = 8, where the initial value of the number K 3 is: 8 . 10 Initial value:

[0388] .

[0389] G3: The first calculation unit 6013 or the remote host 600 checks whether K 2b contains 0. If it contains 0, add 1 to K 10 and return to G2 for recalculation; if it does not contain 0, proceed to G4.

[0390] G4: Calculate K 10 -K 0 +1, and compare it with the decimal number k entered by the user 10 If K 10 -K 0 +1 = k 10 , then convert K 10 to binary code K 2 , and write it into the second storage unit 6022 according to the method of steps A1 to A7; if K 10 -K 0 +1 ≠ k 10 , then add 1 to K 10 and return to G2 for recalculation.

[0391] The essence of the above method is to find the k-th code K without 0 in the b binary code, and use k as the mapping of the binary code. While satisfying the target lamp coding rule, it also ensures the continuity of the number. 2b .

[0392] In this embodiment, converting the target code to the target number includes:

[0393] H1: According to B1 to B3, read the binary code K in the second storage unit 6022 by the handheld terminal 601 or the remote host 600 2 .

[0394] H2: Convert the binary code K 2 to binary code K b in base 2 2-2b by the first calculation unit 6013 or the remote host 600.

[0395] H3: Convert the decimal code K 10 to binary code K b in base 2 2b by the first calculation unit 6013 or the remote host 600, where K 10 The initial value of K 0 is calculated the same as G2.

[0396] H4: Compare K 2-2b with K 2b by the first calculation unit 6013 or the remote host 600. If K 2b = K 2-2b , then output K 10 -K 0 +1 as the decimal number of the binary code K 2 ; if K 2b ≠ K 2-2b , then K 10Return to H3 for recalculation after adding 1.

[0397] The essence of the above method is that at 2 b in the binary encoding, given that its encoding is K 2b , find out which non - zero - containing encoding it is. Its advantage is that the mutual conversion between numbering and encoding can be completed only by using the system's RAM, without occupying the space of the remote host 600, the handheld terminal 601, and the second storage unit 6022 of the target.

[0398] For the convenience of understanding the above method, a certain section of the numbering and encoding of the target with three target lights and three wavelengths is given in Table 2 for reference. Table 2 is: A certain section of the numbering and encoding of the target with three target lights and three wavelengths.

[0399]

[0400] As can be seen from the above table, converting decimal to 2 b in the binary system actually simplifies the judgment of multiple sub - codes in the binary system to the judgment of 1 encoding, and the number N of non - encodable encodings can also be calculated as:

[0401] .

[0402] The foregoing encoding method is based on the forward - facing target light as a benchmark and performs input and decoding in the order from left to right; for the backward - facing target light, since the wavelength combination it outputs is the same as that of the corresponding forward - facing target light, the front - back reverse influence is not considered during input and decoding. The process of reversing the backward - facing target light encoding is completed through the image decoding stage.

[0403] Embodiment of the decoding method for the target light image:

[0404] A decoding method for the target light image, which is based on the above - mentioned equal - light - intensity and parallax - free synchronous image splitting and multi - band imaging device embodiment to obtain the target light image set of the target cluster composed of the above - mentioned target and target light embodiments, and decodes the target light image set, specifically including the following steps:

[0405] 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.

[0406] R2. Screen 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.

[0407] R3. Classify the target lights according to the area of the target light image, and group the target lights with similar areas into the same target.

[0408] In this embodiment, when the imaging device collects the target image within the visual field, its viewing direction is not fixed. If the viewing direction is opposite, the collected target image is also reversed. If not distinguished, it will cause recognition errors according to the aforementioned coding rules. Therefore, the aforementioned target lamp is designed with inconsistent front and back characteristics (circular light spot in the front and annular light spot in the back) to facilitate distinguishing the viewing direction of the imaging device from the image.

[0409] Further, during the execution of step R1, since the target will inevitably be blocked front and back, resulting in the problem that the target lamp light spot is not fully imaged. At this time, the light spot does not present an elliptical shape, but a trimmed edge situation, which we call a residual image. According to the spatial position relationship of the target, the situation of the target image being blocked is mainly divided into the following three types: The first is corner occlusion: mainly manifested as the light spot of the target image being blocked by the corner of another target, presenting an overall concave polygon structure; the second is most occlusion: mainly manifested as most of the light spot of the target image being missing, only showing a crescent-shaped light spot structure; the third is small part occlusion: mainly manifested as very little of the light spot of the target image being missing, only showing a convex crescent-shaped light spot structure.

[0410] Due to the different shapes of the front and back light spots (circular in the front and annular in the back), and the influence of the aforementioned occlusion, it will interfere with the judgment of the light spot type (circular or annular). Whether it is a front or back light spot, due to the lack of background reference, the above several residual images will interfere with the subsequent analysis of the light spot. Therefore, they must be excluded from the recognized light spots and not participate in the analysis. Therefore, it is necessary to ensure that the light spot is complete, and then determine the viewing direction by determining whether it is circular or annular. The set process is as follows: First, define an index (variation degree) that can describe the difference between the residual image and the complete target image, select a target image with the smallest variation degree from the image (although there are occluded situations for the target image, after all, they are in the minority, and there must be unoccluded ones in any image, so the target image with the smallest variation degree must be complete) for target type analysis (circular or annular), so as to complete the orientation of the viewing direction. Therefore, the specific steps of R1 are as follows:

[0411] I1: Perform binarization processing on the front and back target lamp images respectively, obtain the outer boundary through Canny edge detection, and get the light spot area; restore the center coordinates of the target lamp.

[0412] I2: Calculate the variation degree E of the light spot in the image R , according to E R To determine the viewing direction.

[0413] I3: For each image, select a light spot with the smallest variation degree for viewing direction discrimination.

[0414] Among them, I1 includes:

[0415] Perform binarization processing on the front and back target lamp images respectively.

[0416] The center of the identified boundary pixel is stored at the boundary point (x Bi , j , y Bi , j ), (x Bi , j , y Bi , j ) represents the coordinates of the j-th boundary point of the i-th light spot; the total area A of the pixels inside the boundary is counted i .

[0417] The coordinates of the center point G (x Gi , j , y Gi , j ) is restored through the centroid of the pixels in the boundary region, (x Gi , j , y Gi , j ) represents the center of the i-th light spot. When calculating the backward target lamp, the central dark area is not considered, where (x Gi , j , y Gi , j ):

[0418] .

[0419] Among them, I2 includes:

[0420] I21: Calculate the distance S between the centroid G of the i-th light spot i and the j-th boundary point: i , j :

[0421] .

[0422] I22: Calculate the radius R of the equivalent circle of the i-th light spot (shown as the dashed circle in Figure 32 ): ei :

[0423] .

[0424] I23: For the i-th light spot, with the equivalent circle radius R ei as a reference, calculate the deviation of S i , j from the equivalent circle radius R ei , and E Ri is the variability of the light spot:

[0425] .

[0426] The smaller the value of E Ri , the lower the degree of contour change caused by occlusion. When the light spot is a standard circle, this value should be 0.

[0427] Among them, I3 includes:

[0428] I31: Starting from the center of the target lamp light spot and limited by the detected outer boundary, along the x and y directions respectively, obtain the distribution of the gray value k of the image, and calculate the corresponding gray sample variance Sx and S y such as Figure 33 shown:

[0429] .

[0430] .

[0431] I32: Construct the target discriminant:

[0432] .

[0433] I33: If Δ = 0, it is determined that the target light in the image is forward; if Δ ≠ 0, it is determined that the target light in the image is backward.

[0434] The above step R1 also includes:

[0435] According to the variability E R Combined with the spot area A i Evaluate the integrity of its spot. The spots that meet the following conditions should be excluded:

[0436] .

[0437] where is the variability threshold of the unobstructed spot.

[0438] Variability calibration mainly considers two aspects: for the images of the same wavelength, the variability situation is related to the size of its spot (i.e., the distance between the target light and the camera). Because the resolution of the image is certain, the smaller the spot area, the stronger the edge serration and the greater the relative variability. On the other hand, there are differences in the imaging areas of sensors of different wavelengths, and the variability thresholds of the images of each wavelength need to be corrected separately. The variability calibration process includes the following steps:

[0439] Calibrate the variability threshold and determine the initial value of the variability threshold:

[0440] Q1: Use the light target arranged on the site as the calibration object, and set all the light targets to λ k wavelength through the remote host 600, where k ≤ b, b is the total number of wavelengths that each target light can emit, and all are lit.

[0441] Q2: Take 1 image at the λ k wavelength, and obtain the spot area A k,i and the spot variability E Rk,i in the image; i ≤ n, n is the total number of spots in the image (at this time, there may be afterimages in each spot, not all are complete spots).

[0442] Q3: Keep E Rk,iThe correspondence with A k,i , sort A k,i into the sequence {A k , n}, and obtain the corresponding spot variability sequence {E Rk,n}.

[0443] Q4: Find the minimum term E Rk,n of the sequence {E Rk,m . The corresponding terms of the sequence {A k , n} are A k , m .

[0444] Q5: Find the maximum term E Rk,n of the sequence {E Rk,p . The corresponding terms of the sequence {A k , n} are A k , p .

[0445] Q6: If the A k , p in Q5 is the minimum in the sequence {A k , n}, then calculate λ k at the wavelength where the area is A k,i for the spot variability threshold E R0(k，i) :

[0446] .

[0447] Q7: If the A k , p in Q5 is not the minimum in the sequence {A k , n}, then remove this item and re - execute Q5.

[0448] To facilitate the understanding of the above process, an example is given as follows:

[0449] For an image at a certain wavelength, the area and variability of each spot are calculated as shown in Table 3. Table 3 shows: The area and variability of each spot calculated for an image at a certain wavelength.

[0450]

[0451] As can be seen from the above table, among them:

[0452] The minimum variability E Rk,2 = 1.802, and its corresponding A k ,2 =270;

[0453] Maximum Variation E Rk,3 =3.520, which corresponds to A k , 3 =151 <min({ A k , n})=145, remove and screen again;

[0454] Maximum Variation E Rk,3 =2.102, which corresponds to A k , 4 =145=min({ A k , n})=145, meets the requirements;

[0455]

[0456] When the spot area is known to be 255, its variation threshold E R0 =1.838.

[0457] The advantage of this method for calculating the variability threshold is that there is no need to set up targets separately outside the field for calibration, and calibration can be achieved directly using targets arranged 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, and the judgment accuracy is higher.

[0458] In the above step R2, the light spots from the same target light in each wavelength image are merged, mainly to solve the problem of combination and matching of target light images of each wavelength. After the target light is processed by the above-mentioned residual image removal, the light spots from the light spots of each wavelength image are screened out to determine which light spots are from the same target light, so that 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.

[0459] Since each imaging element of the multi-wave imaging camera has the same image plane coordinate system, the multi-band images will be superimposed when each band image is projected onto the virtual imaging plane (the virtual imaging plane is only imaginary for ease of understanding, and each wavelength of light will not overlap and form a secondary image). Figure 34As shown in the figure, taking the three-wavelength target lamp as an example, when all three-wavelength lamp beads are lit, the three images will overlap on the virtual imaging plane, and the combination of the remaining number of lamp beads can be deduced by analogy. To achieve the above overlapping process, it is necessary to first establish a mapping relationship for the spot data of each band image, as Figure 35 shown, and its specific steps include:

[0460] J1: Obtain the outer boundary and the target center coordinates of each wavelength target lamp image, and separately count the number of spots in each wavelength image. The central dark area of the rear target lamp is not considered during the analysis.

[0461] J2: For a target lamp with b wavelengths, taking the wavelength image with the largest number of spots as the reference, its wavelength is λ t , and calculate the distance between the center of each spot image of this wavelength and the center of the spots of other wavelength images respectively:

[0462] .

[0463] Among them, x k represents the abscissa of the k-th spot center in the wavelength image with the largest number of spots; y k represents the ordinate of the k-th spot center in the wavelength image with the largest number of spots; x λc,n represents the abscissa of the n-th spot center in the wavelength image with wavelength λ c ; y λc,n represents the ordinate of the n-th spot center in the wavelength image with wavelength λ c ; D(k, λ c , n) represents the center distance between the k-th spot in the wavelength image with the largest number of spots and the n-th spot in the wavelength image with wavelength λ c .

[0464] J3: Among the numerous points with wavelength λ c , when a certain point can make the value of the previously calculated D(k, λ c , n) the smallest, then the distance between this point and the k-th point with the largest number of spots and wavelength λ t achieves overlap; using D(k, λ c , n), establish the corresponding relationship between each target point with wavelength λ t and the target points of other wavelengths according to the following formula:

[0465] .

[0466] Among them, P(λ t , k) represents the k-th point with the largest number of spots and wavelength λ t ; P(λ c , q) represents the q-th point with wavelength λ c , and this point is related to P(λt , k) overlap.

[0467] J4: Merge the spot centers determined to be the same target lamp.

[0468] The merging processes of the foregoing J1 to J3 only complete the classification of the spot images of each wavelength. In the actual imaging process, due to factors such as the difference in the photosensitive ability of the sensor to different wavelengths, for the same target lamp, it is impossible to achieve complete overlap of the images of different wavelengths in terms of spatial position and shape, and there will always be deviations in position and shape (as shown in Figure 36 -(a)). However, when using the multi-band images of the target lamp to overlap and determine the wavelength combination and spot center of the target lamp, we hope that the centers of the spots of each wavelength coincide. Therefore, it is necessary to merge the spot centers determined to be the same target lamp in J3. For the same target lamp with three wavelengths, using the merging method in J3, there are at most three image merging categories, namely: the first category, single wavelength, no need to merge; the second category, merging of two-wavelength images; the third category, merging of three-wavelength images. However, the so-called merging does not simply coincide these points, but constructs an algorithm to equivalently merge each point into an optimal point. The characteristics that the optimal point after merging should possess are: the sum of the distances from the optimal point to each known point is the smallest. Based on this principle, the specific merging method in step J4 is as follows:

[0469] J41: Equivalently merge the spot centers of different wavelengths of the same target lamp.

[0470] J42: Merge the spot areas of different wavelengths of the same target lamp.

[0471] Among them, J41 includes:

[0472] J411: Set the analysis accuracy: When calculating according to J412 to J418, when the search center gradually approaches the search center M point, the step size change will tend to 0; considering the discreteness of digital imaging, use Δ≤δ as the cut-off condition, where Δ is the step size and δ is the pixel size of the imaging sensor.

[0473] J412: Determine the initial search area: According to the center points P λ1 ~P λm (1≤m≤b) coordinates of the spot centers of different wavelengths of the same target lamp to determine the initial search range, and take min(x i )≤x≤max(x i ), min(y i )≤y≤max(y i ) (1≤i≤m) area as the initial area;

[0474] J413: Determine the initial search center: The geometric center G formed by all the points in the area determined by J412 is used as the initial search center point M, and the coordinates are (asFigure 36 -(as shown in (a)):

[0475] .

[0476] J414: Determine the initial search step: Take half of the smaller value in the height and width directions of the area determined by J412 as the initial search step Δ (as Figure 36 -(as shown in (b)):

[0477] .

[0478] J415: Search area: With the optimal point M as the center and Δ as the step, arrange search points in 8 orthogonal directions around it. For the first calculation, the M point is the G point; the coordinates of each search point are expressed as (as Figure 36 -(as shown in (b)):

[0479] .

[0480] J416: Target calculation: For each search point arranged according to J415, calculate the distance between it and the center points of the light spots of m different wavelengths of the same target lamp respectively and sum them up (as Figure 36 -(as shown in (c)):

[0481] .

[0482] J417: According to J411, perform accuracy judgment. If it is satisfied, output the point position M, which is the optimal point; otherwise, adjust the search center and step according to J418 (as Figure 36 -(as shown in (d));

[0483] J418: For the point position M that does not meet the conditions of J417, consider the following two situations:

[0484] Situation 1: If M coincides with the center point M in J415, it indicates that the initial search area is too large. At this time, update the step Δ in step J415 to Δ / 2 to further narrow the range; then recalculate according to steps J416 - J417 (as Figure 36 -(as shown in (d)).

[0485] Situation 2: If M does not coincide with the center point M in J415, at this time, update the position of the M point. After updating the M point in step J415 to this point, execute steps J416 - J417 (as Figure 36 -(as shown in (e)).

[0486] J419: Process points P λ1 ~P λm according to steps J411 - J418 to complete the merging of the centers of the light spots of different wavelengths of the same target lamp.

[0487] In the above steps, we have completed the equivalent merging of the spot centers of different wavelengths of the same target lamp into one point (i.e., position merging). Next, we continue to merge the spot areas of different wavelengths of the same target lamp (shape merging), so that the superposition of multiple-band images is completely completed. The area merging of multi-band images is not simply to add the areas of each image and then calculate the average value. The weight of the position distribution deviation should be considered, and the areas are weighted and averaged. Based on this principle, in step J4, the specific merging method step J42 includes:

[0488] J421: For the spots of different wavelengths of the same target lamp, calculate its merging center M(x M , y M ) according to J411~J418, and then calculate the square S i of the distance between the merging center M and the spot center of each wavelength:

[0489] ;

[0490] J422: According to the minimum distance deviation D calculated in J416, calculate the distance deviation weight β between the spot centers of different wavelengths of the same target lamp and the merging center M:

[0491] ;

[0492] J423: According to the spot area A i of different wavelengths of the same target lamp, considering the image of the distance deviation weight in J422, calculate the weighted average value of the superposition of the spot areas of each wavelength:

[0493] ;

[0494] J424: Process the spot areas corresponding to points P λ1 ~P λm according to steps J421~J423 to complete the merging of the spot areas of different wavelengths of the same target lamp.

[0495] The above-mentioned steps have completed the recognition of the target image, the elimination of afterimages, and the merging of homologous target images. However, these are only for the images of one target lamp. There are several target lamps on one target. Therefore, it is necessary to classify the numerous target images processed above, and classify the target lamp images on the same target to facilitate the determination of the spatial position of the target. Since the target lamps on the same target are in the same plane and the distances from each target lamp to the camera are very different, the aberration of their respective spot areas is very small; however, due to the scale change of the spatial distribution of different targets being much larger than the distances from each target lamp to the camera, the images can be classified by area, and those with similar areas are classified as the same target. Based on this principle, the specific target lamp clustering method in step R3 is as follows:

[0496] R31: There are data after merging q homologous target lights. The merging center of each homologous target image is (x Mi , y Mi ), and the merging area is A Ci (i ≤ q). When analyzing, the central dark area of the rear target light is not considered;

[0497] R32: Sort A Ci in ascending order;

[0498] R33: Starting from A C1 , compare A Ci with A Ci+1 and calculate the relative deviation value δ i :

[0499] ;

[0500] R34: If δ i > δ 0 , then classify A Ci+1 and A Ci into the same type of target, denoted as the r-th type;

[0501] Otherwise, classify A Ci+1 into the next type of target, denoted as the (r + 1)-th type;

[0502] where δ 0 is the area resolution;

[0503] R35: According to the above steps R31 - R34, assume that there are a total of t types of targets. Then discard the last group of grouped targets, that is, discard the t-th group, and retain the first t - 1 groups as the result of target clustering grouping (as Figure 37 shown, the area change of the ninth type cannot be distinguished, so only take the 1st - 8th type of targets. Since the ninth type cannot be distinguished as the belonging target, it is discarded).

[0504] The aforementioned area resolution δ 0 belongs to the characteristics of the light source itself and needs to be calibrated to determine its initial value. The specific calibration method is as follows:

[0505] Calibrate the area resolution and determine the initial value of the area resolution:

[0506] P1: Use the light targets arranged on the site as the calibration object, and set all the light targets to the wavelength of λ k through the remote host 600, where k ≤ b, and b is the total number of wavelengths that each target light can emit, and all are lit up.

[0507] P2: Take an image under the wavelength of λ k , and obtain the areas A k,i of each light spot in the image, where i ≤ n, and n is the total number of light spots in the image.

[0508] P3: Sort A k,i in descending order to obtain the sequence {A k , n}, where n is the total number of light spots in the image. Calculate the difference Δ k,j = A k , j+1 - A k , j ; j ≤ n, where n is the total number of light spots in the image (there may be afterimages in each light spot at this time, not all are complete light spots).

[0509] P4: Take δ k , 0= that is min(Δ k,j ) and δ k , 0 > 0 as the area resolution of the image at this wavelength.

[0510] P5: Calculate the δ k , 0 values of each light spot image at the b - type wavelength according to P1 - P4 respectively.

[0511] P6: Calculate the variability values of each light spot image at the b - type wavelength respectively, and select the light spot with the smallest variability value as the calculation object.

[0512] P7: Take one light spot with the smallest variability value of each type of wavelength as the calculation object, and calculate the corresponding deviation weight β k for each wavelength of this light spot respectively.

[0513] P8: Weight - average the δ k , 0 values calculated in P5 according to the β k calculated in P7 to obtain the area resolution δ 0 :

[0514] .

[0515] The advantage of calculating and determining the area resolution threshold by this method is that there is no need to separately deploy a target off - site for calibration, and the calibration can be directly achieved by using the targets arranged on - site; because the merged area of the same - source different - wave target images is used for light spot clustering, this resolution threshold takes into account the influence of area differences in imaging of different bands, and the discrimination accuracy is higher.

[0516] After completing the above decoding steps, the method of this embodiment further includes (refer to Figure 38 ):

[0517] Integrate the results of decoding the target lamp image set into image labels to obtain the first data, which specifically includes the following steps:

[0518] V1: Record the time stamp and camera number when the image is taken.

[0519] V2: Determine the viewing direction of the image and screen available target images, using code 1 to represent forward and 0 to represent backward.

[0520] V3: Distinguish different target lights according to the wavelength combination.

[0521] V4: Complete the coordinates of each target light image in the image plane, and record the position of each target light as (x i , y i ).

[0522] V5: Find all the target lights included in each target, and store the positions of the target lights in the same group as a set.

[0523] V6: Generate a coding sequence using the set in V5 to make each target light image have a unique identifiable number.

[0524] V6 includes:

[0525] V61: If the determination in V2 is 1, for each set in V5, sort the x coordinates of the target lights it contains in ascending order; for example, if a target light set is {(12, 2), (5, 5), (8, 9)}, sort it in x ascending order of coordinates, and the sorted sequence is: (5, 5), (8, 9), (12, 2).

[0526] If the determination in V2 is 0, for each set in V5, sort the x coordinates of the target lights it contains in descending order; for example, if a target light set is {(12, 2), (5, 5), (8, 9)}, sort it in x descending order of coordinates, and the sorted sequence is: (12, 2), (8, 9), (5, 5).

[0527] V62: Replace the coordinates corresponding to the target light sequence in V61 with the corresponding target light codes in V3 to obtain a new sequence, which is the entire target number. For example, if the known target light sequence is (12, 2), (8, 9), (5, 5), and its corresponding target light codes are 001, 110, 101, then the number of this target is 001110101.

[0528] The above method correlates and integrates the target light direction, target light image coordinates, and the information of the target number to which they belong without referring to anything outside a single image, realizing the autonomous identification of target images.

[0529] Furthermore, refer to Figure 39As shown in the figure, to complete the autonomous identification of the target, the following steps are also included:

[0530] N1: When the camera is triggered to collect images, a single-channel image file will be generated separately for the target lamp images of each wavelength in the camera and transmitted to the remote host 600 through in-field communication.

[0531] N2: The remote host 600 calculates and stores the image tags.

[0532] N3: Generate a visualized composite image:

[0533] N4: The remote host 600 stores the "Unix timestamp", "camera number", and the "viewing direction", "band number", "target image code", and "target image center" obtained in N2 as a data file DATA.

[0534] N5: Integrate the images of each band, the visualized composite image, and the data file DATA into one TIFF file.

[0535] Among them, step N3 includes:

[0536] N31: Generate a blank image with the same resolution as the images of each band and the color level of each pixel channel being 0.

[0537] N32: Traverse the images of each band according to the following rules and assign values to the color levels of each pixel in the blank image generated in N31:

[0538] For any pixel (i, j) of the images of each band in the image plane coordinate system, if the color level value of any one of its images is 255, then assign the color level of 255 to the pixel (i, j) at the corresponding position in the blank image generated in N31; otherwise, assign 0.

[0539] For easy understanding of the above process, an example is given as follows. For example, the tag of an image is:

[0540] (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)

[0541] The annotation is shown as follows:

[0542] Unix timestamp: 1694617868

[0543] It means that the image was taken at 23:11:08 on September 13, 2023.

[0544] Viewing direction: 01

[0545] Indicates that the current viewing direction is forward.

[0546] Target image code: 100110001, 100110010, 100110011

[0547] Indicates that there are 3 targets in total, namely 233, 234, and 235, in the image. Among them, the states of the three lights of target 233 are λ 1, λ 1 + λ 2, λ 3; the states of the three lights of target 234 are λ 1, λ 1 + λ 2, λ 2; the states of the three lights of target 235 are λ 1, λ 1 + λ 2, λ 2 + λ 3.

[0548] Target image center:

[0549] 491.880,514.255 , 906.287, 675.024, 1320.694,835.793 , 2548.923, 2550.323, 2806.690,2456.659 , 3064.456, 2362.994

[0550] Indicates target 233 λ 1, λ 1 + λ 2, λ 3. The corresponding center coordinates of the three target lights of the image are (491.880, 514.255), (906.287, 675.024), and (1320.694, 835.793) respectively.

[0551] Indicates that the corresponding center coordinates of the three target lights, namely λ1, λ1 + λ2, and λ2, of target 234 are (2548.923, 2550.323), (2806.690, 2456.659), and (3064.456, 2362.994) respectively.

[0552] Through a TIFF file, that is, integrating the data information of each band image and its joint processing, this data processing method can effectively solve the problems of large number of image acquisitions by each camera and difficult storage and classification. Moreover, since it contains all the information starting from the original image, its traceability is strong and it is convenient for later verification and proofreading.

[0553] Embodiment of the Spatial Multi-Target Image Autonomous Identification System:

[0554] A spatial multi-target image autonomous identification system, comprising:

[0555] The above-mentioned target, the above-mentioned imaging device, and the above-mentioned coding device; the number of arranged targets is a first preset number, and they are arranged staggeredly along a preset baseline, and the display directions of the target lights with the same shape on all targets are the same; the target lights of adjacent targets on the same baseline are staggeredly distributed in the shooting direction of the imaging device; the number of arranged imaging devices is a second preset number, which is used to collect the target light images on the targets; the coding device is used to set the on-off combination of the target lights, encode the targets, and decode based on the target light images collected by the imaging device to complete autonomous identification.

[0556] Embodiment of the Spatial Multi-Target Image Autonomous Identification Method:

[0557] A spatial multi-target image autonomous identification method, based on the above-mentioned spatial multi-target image autonomous identification system, the method comprises:

[0558] U1. Arrange the target cluster: Arrange a first preset number of targets staggeredly along a preset baseline, and the display directions of the target lights with the same shape on all targets are the same; the target lights of adjacent targets on the same baseline are staggeredly distributed in the shooting direction of the imaging device.

[0559] U2. Target coding: Set the on-off combination of the target lights on each target through the coding device to encode the targets.

[0560] U3. Collect target light images and decode: Use the imaging device to collect the target light images on the targets to obtain a target light image set; decode the target light image set to obtain decoded data, and the decoded data includes the target light direction, the target light image coordinates, and the number information of the target to which the target light belongs.

[0561] U4. Autonomous identification: Associate the decoded data with the target light image set to complete autonomous identification.

[0562] In U1, there are more than two preset baselines, and the preset baseline is a continuous space curve on any wall surface in the application scenario.

[0563] Although the present invention has been disclosed above with preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make many possible changes and modifications to the technical solution of the present invention, or modify it into an equivalent embodiment with equivalent changes, without departing from the scope of the technical solution of the present invention. 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 for autonomous image identification, characterized in that: The target rod (100) comprises a target base (300) and a target light (200). One end of the target rod (100) is adjustably mounted on the target base (300). The target light (200) is mounted on the target rod (100) and can display light on the front and rear surfaces of the target rod (100).

2. The target for autonomous image identification according to claim 1, characterized in that: The target base (300) is provided with a base frame (303), and 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 mounting the fixing fastener (3031), and leveling screw holes (3034) are respectively provided on both sides of the mounting through hole (3033) for mounting the pair of leveling bolts (3032).

3. The target for autonomous image identification according to claim 2, characterized in that: The bottom of the leveling bolt (3032) is spherical.

4. The target for autonomous image marking according to claim 3, characterized in that: The base frame (303) is provided with an arc guide groove (301); the target rod (100) is hinged to the target seat (300), and the hinge point is located at the center of the arc guide groove (301); 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) at a specified position.

5. The target for autonomous image marking according to any one of claims 1 to 4, characterized in that: Three target lights (200) are provided, and the three target lights (200) are evenly spaced along the target pole (100), and the lights displayed by the three target lights (200) on the front and rear surfaces are the same in the same direction and different in different directions.

6. The target for autonomous image marking according to any one of claims 1 to 4, characterized in that: The surface of the target rod (100) is provided with a low-reflectivity matte coating.

7. A target light (200) for a target for autonomous image identification as claimed in any one of claims 1 to 6, characterized in that: The invention comprises a rear scattering cover (201), a rear hyperbolic reflection cavity (202), a reflection cone (203), a lamp bead (204), a front hyperbolic reflection cavity (205), and a front scattering cover (206); the rear hyperbolic reflection cavity (202) is butt-jointed with the front hyperbolic reflection cavity (205) and the reflection cone (203) is clamped and fixed inside; a plurality of lamp beads (204) are provided and are dispersedly arranged at the front and rear 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).

8. The target light (200) according to claim 7, characterized in that: A light shielding hole is provided in the centre of the front scattering cover (206) or the rear scattering cover (201); 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.

9. The target light (200) according to claim 8, characterized in that: The lamp beads (204) include LED lamp beads of multiple different wavelengths, and LED lamp beads of multiple different wavelengths are arranged on both the front and rear sides of the reflection cone (203), and the LED lamp beads of multiple different wavelengths on the same side of the reflection cone (203) are arranged in a staggered manner.

10. The target light (200) according to claim 9, characterized in that: The lamp beads (204) on the front and rear sides of the reflection cone (203) are arranged in a staggered manner.

11. The target light (200) according to claim 10, characterized in that: The optical centers of the lamp beads (204) on the front and rear sides of the reflection cone (203) are located on the same plane.

12. The target light (200) according to claim 11, characterized in that: The rear hyperbolic reflection cavity (202), the front hyperbolic reflection cavity (205) and the reflection cone (203) are coaxially arranged.

13. A method for fixing a target for image autonomous marking as claimed in any one of claims 1 to 6 on an uneven surface, characterized in that: The following steps are involved: S1: Pre-installation, passing the fixing fastener (3031) through the installation through hole (3033) on the target base (300) and driving it into the uneven surface, and making the fixing fastener (3031) not completely press the target base (300); S2: Leveling, screwing the leveling bolts (3032) into the leveling screw holes (3034) so ​​that the spherical parts at the bottom of the leveling bolts (3032) contact and abut against the uneven surface, and adjusting the screwing depth of the leveling bolts (3032) so that the posture of the target base (300) tends to the target installation posture; S3: Fixing. When the target base (300) reaches the target installation posture, tighten the fixing fastener (3031) so that the target base (300) is completely fixed, thereby completing the fixed installation of the target on the uneven surface.

14. A spatial multi-target image autonomous identification system, characterized in that: include: The target according to any one of claims 1 to 6; Imaging device; Coding device; The target is equipped with a target light as claimed in any one of claims 7 to 12, the number of the targets being arranged is a first preset number, and being staggered along a preset baseline, the target lights (200) on all targets having the same shape have the same display direction of light; the target lights (200) on adjacent targets on the same baseline are staggered in the shooting direction of the imaging device; The number of the imaging devices arranged is a second preset number, and is used to collect the target light image on the target; The encoding device is used to set the on / off combination of the target light (200), encode the target, and decode the target light image captured by the imaging device to complete autonomous identification.

15. A spatial multi-target image autonomous identification method, based on the spatial multi-target image autonomous identification system according to claim 14, characterized in that: The method comprises: U1. Arranging a target cluster: arranging a first preset number of targets in a staggered manner along a preset baseline, wherein target lights (200) of the same shape on all targets have the same display direction of light; and target lights (200) of adjacent targets on the same baseline are staggered in the shooting direction of the imaging device; U2, target coding: the coding device is used to set the on and off combination of the target lights (200) on each target, thereby coding the target; U3, collecting and decoding target light images: using the imaging device to collect the target light images on the target to obtain a target light image set; decoding the target light image set to obtain decoded data, the decoded data including the target light direction, the target light image coordinates, and the number information of the target to which the target light belongs; U4, autonomous identification: associate the decoded data with the target light image set to complete autonomous identification.

16. The spatial multi-target image autonomous identification method according to claim 15, characterized in that: In U1, there are more than two preset baselines, and the preset baselines are continuous space curves on any wall surface in the application scene.

Images