Long and short chain accurate positioning method and system based on image recognition

Abstract

The invention discloses a long-chain and short-chain accurate positioning method and system based on image recognition, and the method comprises the steps: employing an image recognition algorithm and a positioning mileage correction algorithm, carrying out the whole-line data collection through a line-scan digital camera, carrying out the precise recognition of an electronic tag and a hectometer tag on a line through the image recognition algorithm, and constructing a basic database of the electronic tag and the hectometer tag. The line number between the two hectometer marks is calculated through the image, and the actual distance between the two hectometer marks is accurately calculated. And the timestamps corresponding to the two hectometer mark pictures are utilized to determine a large interval range to which the current hectometer mark data belong. Meanwhile, kilometer post information generated by the photoelectric encoder can be retrieved, the accurate range of the current kilometer post is judged, and the recognition accuracy of the hectometer post is further verified. Therefore, the system can determine the current actual kilometers and calculate the end point of the actual kilometers in combination with the actual distance, so that the automatic error correction function of the long and short chains is realized. Therefore, the troubleshooting process of potential problems is greatly simplified, and the efficiency of maintenance operation is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of train track maintenance, and in particular to a method and system for accurately locating long and short chains based on image recognition. Background Art

[0002] The phenomenon of long and short chains in subways refers to the situation in which the actual line mileage deviates from the preset mileage due to specific factors (such as local line adjustments, segmented measurements, etc.) during the design and construction process of subway lines. When the preset 100-meter mark mileage exceeds the actual ground mileage, this phenomenon is defined as a short chain; otherwise, it is called a long chain.

[0003] The phenomenon of long and short subway chains mainly stems from the following aspects:

[0004] A. Segmented measurement: During the field survey of subway lines, due to technical limitations and practical operational requirements, the measurement work is usually divided into multiple parts. Therefore, there may be a mismatch between the starting and ending mileages of each segment measurement, which may lead to a broken chain and the appearance of a long and short chain.

[0005] B. Local line adjustment: During the design and construction of subway lines, local adjustments may be required due to factors such as terrain and environment. Such adjustments often cause deviations between the original 100-meter mark and the actual mileage, thus forming a long-short chain.

[0006] C. Measurement error: During the measurement process, the 100-meter mark may not match the actual mileage due to calculation errors or instrument accuracy problems. This is also an important reason for the generation of long and short chains.

[0007] The phenomenon of long and short chains in subways will have a certain impact on the construction and operation of the line. Therefore, the phenomenon of long and short chains needs to be properly handled during the design and construction process, and is generally corrected through manual data adjustment. In the daily line operation and maintenance process, the line data needs to be reviewed regularly, and the necessary data of long and short chains needs to be sorted out according to the actual situation.

[0008] Although the long-short chain of subways is an unavoidable phenomenon in the design and construction of subway lines, and its causes are complex and diverse, it has a certain impact on the construction and operation of subways. However, in subway projects, the possibility of long-short chain still needs to be fully considered, and corresponding measures need to be taken to deal with and adjust it.

[0009] Although the needs of daily operation and maintenance can be met by manually organizing and updating long and short chain data on a regular basis, this process requires frequent data verification and calibration. Especially when it involves unified data management of multiple operating systems, there may be problems such as data confusion and inaccurate mileage positioning, which increases the difficulty of troubleshooting, prolongs the troubleshooting cycle, and consumes a lot of energy. Summary of the invention

[0010] In view of this, the present invention provides a method and system for accurately locating long and short chains based on image recognition to solve the above problems.

[0011] In order to solve the above technical problems, the present invention provides a method for accurately locating long and short chains based on image recognition, comprising:

[0012] After entering the first and second endpoint information of the on-site operation line segment and installing 100-meter markers at preset intervals, a 100-meter marker basic database and an electronic tag basic database are constructed; the train and the collection equipment are started to collect the movement information of the train on the operation line segment; the 100-meter markers and / or electronic tags are used to obtain the real-time mileage information of the train;

[0013] The real-time odometer of the train is generated by collecting electronic tag information through the tag reader and obtaining TLL signal counts through the photoelectric encoder; the real-time odometer is corrected by collecting the real-time mileage information of the train, and a real-time tag positioning difference odometer and a real-time line information basic database are generated;

[0014] The TLL signal is sent to the camera component through signal frequency doubling, and after conversion, a tunnel image and / or a roadbed image is generated to determine whether a hundred-meter mark or an electronic tag is recognized in any image. If so, the real-time calculation of the hundred-meter mark and / or the electronic tag is started to obtain real-time mileage information and update the real-time line information basic database;

[0015] The real-time route information basic database is corrected through the real-time calculation results of the hundred-meter mark and / or electronic tag; wherein,

[0016] Based on the calculation results of the electronic tags, the long and short chains of the electronic tags are generated; based on the calculation results of the hundred-meter mark, the long and short chains of the hundred-meter mark are generated;

[0017] Calculate the difference between the length and short chain of each hundred-meter mark and / or electronic tag, and generate a statistical list of the length and short chain of the operation line segment based on the collection results of the first and second endpoints of the train passing through.

[0018] As an optional method, the electronic tag information is collected through the tag reader to generate the real-time odometer of the train, including:

[0019] Start / wait for the tag reader to determine whether the tag information is read in real time;

[0020] If yes, query the basic data to obtain real-time mileage information;

[0021] If not, start / wait for the tag reader to form an electronic tag basic database, obtain real-time mileage information, form a real-time mileage information table, perform real-time corrections based on the real-time mileage information, and continue to determine whether the tag information is read in real time.

[0022] As an optional method, the TLL signal count is obtained through the photoelectric encoder to generate the real-time odometer of the train, including:

[0023] The TTL signal count is obtained from the photoelectric encoder and given to the camera component through the signal multiplication TTL signal. The mileage information is currently accumulated through the TTL signal to obtain the real-time mileage positioning information and form a real-time line information basic database. The real-time line information basic database is added to form a real-time mileage information table. Real-time correction is performed through the real-time mileage information table to generate a real-time tag positioning difference mileage table. The real-time mileage positioning information is obtained synchronously and the real-time line information basic database is updated. The electronic tag long and short chain is formed through the average value of the tag long and short chain and the correction difference of the electronic tag basic database.

[0024] As an optional method, real-time calculation of the 100-meter mark includes:

[0025] Waiting / starting the calculation of hundred-meter mark recognition, calculating the first mileage of the hundred-meter mark through sampling data, obtaining the real-time second mileage of the hundred-meter mark through the photoelectric encoder, obtaining the hundred-meter mark unique code generated by the unique identifier through the average value of the first mileage of the hundred-meter mark and the second mileage of the hundred-meter mark, and comparing the hundred-meter mark difference with the hundred-meter mark basic database, and judging whether the minimum preset difference is obtained;

[0026] If so, automatically record the current section 100-meter mark length and short chain;

[0027] If not, compare the average value of the first mileage of the hundred-meter mark and the second mileage of the hundred-meter mark with the hundred-meter mark difference in the basic database, wait / start the hundred-meter mark recognition calculation, and form the hundred-meter mark long and short chain through the average value of the label long and short chain and the corrected difference in the hundred-meter mark basic database.

[0028] As an optional method, electronic tag real-time calculation includes:

[0029] Waiting for / starting electronic tag identification calculation, calculating the first mileage of the electronic tag through sampling data, obtaining the real-time second mileage of the electronic tag through the photoelectric encoder, automatically generating the electronic tag unique code by comparing the average value of the first mileage of the electronic tag and the second mileage of the electronic tag with the electronic tag difference in the electronic tag basic database, and judging whether the minimum difference is found;

[0030] If yes, automatically record the length of the electronic tag chain in the current section;

[0031] If not, compare the average of the first mileage of the electronic tag and the second mileage of the electronic tag with the electronic tag difference in the electronic tag basic database, wait / start the electronic tag identification calculation, and form the electronic tag long and short chain through the average of the tag long and short chain in the electronic tag basic database and the corrected difference.

[0032] As an optional method, the difference between the length and short chain of each hundred-meter mark and / or electronic tag is calculated, and a statistical list of the length and short chain of the operation line segment is generated according to the collection results of the first end point and the second end point of the train passing, including:

[0033] Correct the current 100-meter mark basic database through the 100-meter mark long and short chain, and automatically add the current 100-meter mark long and short chain to the dynamic long and short chain basic database;

[0034] The electronic tag long-short chain formed by the average value of the electronic tag long-short chain and the difference value amends the current electronic tag basic database, and automatically adds the current electronic tag long-short chain to the dynamic long-short chain basic database;

[0035] Determine whether the train departs from the first endpoint and arrives at the second endpoint. If so, end; if not, correct the current basic database through the long and short chains of the hundred-meter mark, calculate the long and short chain difference between each hundred-meter mark and / or electronic tag, and generate an automatic statistical list of long and short chains.

[0036] As an optional method, electronic tags are installed at 100-meter points throughout the entire kilometer of the steel track bed to record and mark the position of each 100-meter mark; the photoelectric encoder is installed on the axle and also includes a linear array camera, which is installed in the middle of the train body and on both sides of the front of the train.

[0037] On the other hand, the present invention also provides a long and short chain precise positioning system based on image recognition, which is used to execute the long and short chain precise positioning method based on image recognition as described above, comprising:

[0038] Hardware equipment system, including electronic tags, photoelectric encoders and high-definition 4K linear array cameras;

[0039] Software system, the software system includes real-time data acquisition system and real-time data analysis system; the real-time data acquisition system is used to collect real-time mileage data of the photoelectric encoder, and use the reader to capture the installed electronic tag information; and drive the linear array camera to capture the line label and hundred-meter mark information to build a complete mileage data line information library.

[0040] The beneficial effects of the present invention are:

[0041] The present invention mainly uses image recognition algorithm and positioning mileage correction algorithm, and collects data along the entire line through 6 high-definition linear array cameras. The image recognition algorithm can accurately identify RFID tags and hundred-meter marks on the line, and then build a basic database of RFID and hundred-meter marks. Subsequently, the number of rows between the two hundred-meter marks is calculated through the image, and the actual distance between the two is accurately calculated. Using the timestamps corresponding to the two hundred-meter mark images, the corresponding RFID accurate interval can be reversely searched to determine the large interval range to which the current hundred-meter mark data belongs. At the same time, the system can also retrieve the kilometer mark information generated by the photoelectric encoder, determine the accurate range of the current kilometer mark, and further verify the recognition accuracy of the hundred-meter mark. If there is no misjudgment, the system can determine the current actual kilometer number, and calculate the actual kilometer number end point in combination with the actual distance, realizing the automatic error correction function of the long and short chains. The significant improvement will greatly simplify the troubleshooting process of potential problems, thereby improving the overall efficiency of maintenance operations. It improves work efficiency and reduces work difficulty. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 A schematic diagram of the main process provided for Example 1 of the present invention;

[0043] Figure 2 A schematic diagram of the system workflow provided by Example 1 of the present invention;

[0044] Figure 3 A schematic diagram of a vehicle-mounted high-definition camera provided in Example 1 of the present invention;

[0045] Figure 4 A schematic diagram of a high-definition camera mounted on the front of a vehicle provided in Example 1 of the present invention;

[0046] Figure 5 This is a statistical table of positioning accuracy of the three algorithms provided in Example 2 of the present invention. DETAILED DESCRIPTION

[0047] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention is further described in detail below in conjunction with specific implementation methods.

[0048] Example 1

[0049] The core of this embodiment is to propose a method and system for accurately locating the long and short chain positions based on image recognition technology. The technical implementation of this embodiment mainly includes three key areas: hardware device system, software system and positioning algorithm.

[0050] The hardware equipment system is mainly composed of the following core components: electronic tags, photoelectric encoders and high-definition 4K linear array cameras (with a line frequency of 60K). Specifically, the electronic tags are accurately installed at the hundred-meter mark of the entire kilometer position in the steel track bed. Its function is to record and mark the mileage position of each hundred-meter mark, form a basic database and integrate it into the positioning system. The photoelectric encoder is fixedly installed on the axle to accurately calculate the mileage and speed of the train. In the middle of the electric passenger car body, this embodiment is equipped with two high-definition 4K linear array cameras, and its field of view design ensures that it can fully cover the entire track bed area (such as Figure 1 In addition, in order to meet the effective range of the 100-meter standard installation, this embodiment also installs two high-definition 4K line array cameras on the left and right sides of the front of the electric passenger car (as shown in Figure 2 shown).

[0051] The core components of the software system include the real-time data acquisition system and the real-time data analysis system. Among them, the real-time data acquisition system focuses on collecting real-time mileage data from the photoelectric encoder, and uses the reader to capture the installed electronic tag information. In addition, the system also captures the route label and 100-meter mark information through 6 linear array cameras (installed on the front and body of the vehicle respectively) to build a complete mileage data route information library. Each piece of data is accompanied by a timestamp to ensure the accuracy and traceability of the data.

[0052] In terms of positioning algorithms, this embodiment mainly uses image recognition algorithms and positioning mileage correction algorithms. Through 6 high-definition linear array cameras to collect data along the entire line, the image recognition algorithm can accurately identify RFID tags and hundred-meter marks on the line, and then build a basic database of RFID and hundred-meter marks. Subsequently, the number of rows between the two hundred-meter marks is calculated through the image, and the actual distance between the two is accurately calculated. Using the timestamps corresponding to the two hundred-meter mark images, this embodiment can reversely search the corresponding RFID accurate interval, thereby determining the large interval range to which the current hundred-meter mark data belongs. At the same time, the system can also retrieve the kilometer mark information generated by the photoelectric encoder, determine the accurate range of the current kilometer mark, and further verify the recognition accuracy of the hundred-meter mark. If there is no misjudgment, the system can determine the current actual kilometres, and calculate the actual kilometre end point in combination with the actual distance, thereby realizing the automatic error correction function of the long and short chains.

[0053] See also Figure 1-Figure 3 , this embodiment is implemented as follows:

[0054] After entering the first and second endpoint information of the on-site operation segment and installing 100-meter markers at preset intervals, a 100-meter marker basic database and an electronic tag basic database are constructed; the train and the acquisition equipment are started to collect the movement information of the train on the operation segment; the 100-meter markers and / or electronic tags are used to obtain the real-time mileage information of the train; the electronic tag information is collected through the tag reader, and the TLL signal count is obtained through the photoelectric encoder to generate the real-time odometer of the train; the real-time mileage information of the train is collected to correct the real-time odometer, and a real-time tag positioning difference odometer and a real-time line information basic database are generated; the TLL signal is sent to the camera component through signal frequency multiplication , after conversion, a tunnel image and / or a track bed image is generated, and it is determined whether a hundred-meter mark or an electronic tag is recognized in any image. If recognized, the real-time calculation of the hundred-meter mark and / or the electronic tag is started, the real-time mileage information is obtained, and the real-time line information basic database is updated; the real-time line information basic database is corrected according to the results of the real-time calculation of the hundred-meter mark and / or the electronic tag; based on the calculation results of the electronic tag, a long and short chain of the electronic tag is generated; based on the calculation results of the hundred-meter mark, a long and short chain of the hundred-meter mark is generated; the difference in the long and short chain between each hundred-meter mark and / or electronic tag is calculated, and a statistical list of the long and short chains of the operation line segment is generated according to the collection results of the first endpoint and the second endpoint of the train.

[0055] Please refer again Figure 2 , Figure 2 The implementation is as follows:

[0056] 1-1 Input the basic information of the AB points of the on-site line, 1-2 Form the basic database of 100-meter mark, 1-3 Form the basic database of electronic tags, 1-4 Automatically generate the unique code of the 100-meter mark / tag, 1-5 Query the basic data to obtain real-time mileage information, 1-6 Determine whether to reach point B, 1-7 No Automatically generate the unique code of the 100-meter mark / tag, 1-8 Yes - End,

[0057] 2-1 Start / wait for the tag reader, 2-2 Whether the tag information is read in real time, 2-3 Yes - query the basic data to obtain the real-time mileage information, 2-4 No - start / wait for the tag reader, 2-5 form the electronic tag basic database, 2-6 obtain the real-time mileage information X, 2-7 form the real-time mileage information table X1, perform real-time correction through X, 2-8 continue to determine whether the tag information is read in real time.

[0058] 3-1 Obtain TLL signal count from the photoelectric encoder, 3-2 Give the camera component through the signal frequency doubling TTL signal, 3-3 Currently accumulate mileage information through TTL signal (single signal is about 263mm), 3-4 Obtain real-time mileage positioning information and form a real-time line information basic database, 3-5 Add the real-time line information basic database to form a real-time mileage information table X1, and make real-time corrections through X, 3-6 Through X1-X=Y1, form a real-time tag positioning difference mileage table, 3-7 Synchronous correction-obtain real-time mileage positioning information and form an updated line information basic database, 3-8 Form a label length chain G through the average value of the label length chain K and the difference Y1,

[0059] 4-1 Give the camera component a TTL signal through the signal frequency doubling signal, 4-2 Form a 1mm TTL signal through the current signal, 4-3 Form a tunnel picture / form a roadbed picture, 4-4 Determine whether the hundred-meter mark / electronic tag is recognized, 4-5 Yes - obtain real-time mileage positioning information and form a real-time line information basic database, 4-6 Wait / start the hundred-meter mark recognition calculation, wait / start the electronic tag recognition calculation, 4-7 No - form a tunnel picture / form a roadbed picture,

[0060] 5-1 Waiting / starting hundred-meter mark identification calculation, 5-2 Calculating mileage D1 through sampling data, 5-3 Obtaining real-time mileage C1 through photoelectric encoder, 5-4 Comparing the average value CD1 of D1 and C1 with the hundred-meter mark difference G1 of the basic database, 5-5 Obtaining the hundred-meter mark / label unique code generated by the unique identifier, 5-6 Whether to find the minimum difference L, 5-7 Yes - automatically recording the current interval hundred-meter mark length chain L, 5-8 No - Comparing the average value CD1 of D1 and C1 with the hundred-meter mark difference G1 of the basic database, 5-9 Waiting / starting hundred-meter mark identification calculation, 5-10 Forming the label length chain G through the average value of the label length chain K and the difference Y1,

[0061] 6-1 Waiting / starting electronic tag identification calculation, 6-2 Calculating mileage E1 through sampling data, 6-3 Obtaining real-time mileage F1 through photoelectric encoder, 6-4 Comparing the average value EF1 of E1 and F1 with the electronic tag difference G1 in the basic database, 6-5 Automatically generating a hundred-meter mark / label unique code, 6-6 Whether to find the minimum difference L, 6-7 Yes - automatically recording the electronic tag length chain K in the current section, 6-8 No - Comparing the average value EF1 of E1 and F1 with the electronic tag difference G1 in the basic database, 6-9 Waiting / starting electronic tag identification calculation, 6-10 Forming the label length chain G through the average value of the label length chain K and the difference Y1,

[0062] 7-1 Correct the current basic database through the hundred-meter mark long and short chain L, 7-2 Automatically add the current L to the dynamic long and short chain basic database, 7-3 Form the label long and short chain G through the average value of the label long and short chain K and the difference Y1, 7-4 Automatically add the current G to the dynamic long and short chain basic database, 7-5 Determine whether point B is reached, 7-6 Yes - end, 7-7 No - Correct the current basic database through the hundred-meter mark long and short chain L, calculate the difference in the length and short chain between each hundred-meter mark / electronic tag, and finally form the latest automatic statistical list of the length and short chain.

[0063] The specific embodiments mentioned above are a preferred form of this embodiment and are not intended to limit it. The scope covered by this embodiment is widely applicable to all mileage calculation technologies based on image recognition electronic tags, hundred-meter markers, and linear array cameras, as well as solutions for long and short chain positioning of rail transit vehicles using coded odometers. These solutions are all within the protection scope of this embodiment.

[0064] Example 2

[0065] In order to verify the accuracy and effectiveness of the algorithm currently used in the above-mentioned embodiment 1, this embodiment has formulated a set of solutions including three different experimental steps after careful consideration and planning, the purpose of which is to fully demonstrate the leading level of the algorithm of this embodiment. While pursuing the accuracy of the verification results and the close correlation with the actual operation, this embodiment selected a railway line with long and short chain problems in actual operation as the experimental site. The specific information of the line is: from the starting point A to the end point B, the up and down lines extend 5 kilometers each, and 48 marking points are evenly arranged every 100 meters along the line. At the beginning of the experiment, the train first departs from point A, moves upward along the line, passes all the marking points in turn, until it reaches point B, then turns around and goes down, and returns to point A from point B. Through such a round-trip process, the entire data detection cycle is completed.

[0066] During the entire data detection process, this embodiment uses specific software tools to derive the result data obtained by three different algorithms: the photoelectric encoder mileage accumulation algorithm, the electronic tag correction positioning algorithm, and the positioning algorithm based on image recognition technology. Subsequently, this embodiment calculates the difference by comparing the mileage information in these result data with the position information in the actual 100-meter mark image, and records in detail the frequency of occurrence of deviation values ​​within the range of less than 1 meter, less than 3 meters, less than 5 meters, and less than 10 meters. Through this series of calculations and analyses, this embodiment is able to examine and evaluate the positioning accuracy of each algorithm from a theoretical perspective. In daily maintenance operations, the improvement of the precise positioning accuracy of long and short chains will significantly simplify the process of troubleshooting potential problems.

[0067] After completing the collection and collation of all data, this embodiment obtains corresponding statistical data. Figure 5 It is presented in a graphical form, which intuitively presents the proportion of each algorithm in different deviation ranges, providing a quantitative reference basis for the algorithm evaluation of this embodiment.

[0068] according to Figure 5 From the data in the table, it can be seen that in the range of error less than 1M, the proportion of algorithm A in this embodiment is only 23.96%. In the process of statistical analysis, this embodiment found that this small error mainly occurs in the first 2KM interval from point A to point B, and the actual long and short chain error in this interval is only 3M. However, once the moving distance increases, this embodiment can clearly see that the positioning accuracy is gradually decreasing. Especially at the position of K4+400M, due to the actual existence of 25M long and short chains, this directly leads to a significant decrease in positioning accuracy.

[0069] For algorithm B, after introducing the per-kilometer electronic tag correction, it effectively avoids the influence of the current whole-kilometer long and short links on positioning. After correction, the proportion of algorithm B in the range of positioning error less than 3M increased from the original 31% to 77%, which is undoubtedly a significant performance improvement. Compared with algorithm A, the overall positioning accuracy of algorithm B has increased by 40%, which is a very significant improvement.

[0070] Let’s look at algorithm C. Compared with algorithm B, after re-positioning, the positioning accuracy of less than 1M is directly improved from 32% to 98%, a full 60% improvement. This improvement fundamentally solves the positioning deviation problem caused by long and short chains.

[0071] Through the above solution, in daily maintenance work, the significant improvement of the long and short chain positioning accuracy will greatly simplify the troubleshooting process of potential problems, thereby improving the overall efficiency of maintenance work. This improvement has a far-reaching impact on the workflow of this embodiment. It not only improves the work efficiency of this embodiment, but also reduces the work difficulty of this embodiment. In general, this improvement is very valuable.

[0072] The above are only preferred embodiments of the present invention. It should be noted that the above preferred embodiments should not be regarded as limiting the present invention, and the protection scope of the present invention should be based on the scope defined by the claims. For ordinary technicians in this technical field, several improvements and modifications can be made without departing from the spirit and scope of the present invention, and these improvements and modifications should also be regarded as the protection scope of the present invention.

Claims

1. A long and short chain precise positioning method based on image recognition, characterized in that: include: Enter the first and second endpoint information of the field operation line segment, and after installing the 100-meter markers at preset intervals, build a 100-meter marker basic database and an electronic tag basic database; Starting the train and the collection equipment to collect the movement information of the train on the operation line segment; the hundred-meter mark and / or electronic tag is used to obtain the real-time mileage information of the train; The real-time odometer of the train is generated by collecting electronic tag information through a tag reader and obtaining TLL signal counts through a photoelectric encoder; the real-time odometer is corrected by collecting the real-time mileage information of the train, and a real-time tag positioning difference odometer and a real-time line information basic database are generated; The TLL signal is sent to the camera component through signal frequency doubling, and a tunnel image and / or a roadbed image is generated after conversion, and it is determined whether a hundred-meter mark or an electronic tag is recognized in any image. If so, real-time calculation of the hundred-meter mark and / or the electronic tag is started, real-time mileage information is obtained, and the real-time line information basic database is updated; The real-time route information basic database is corrected through the real-time calculation results of the hundred-meter mark and / or electronic tag; wherein, Based on the calculation results of the electronic tags, the long and short chains of the electronic tags are generated; based on the calculation results of the hundred-meter mark, the long and short chains of the hundred-meter mark are generated; Calculate the difference between the length and short chain of each hundred-meter mark and / or electronic tag, and generate a statistical list of the length and short chain of the operation line segment based on the collection results of the first and second endpoints of the train passing through.

2. According to the method of claim 1, the method is characterized in that: The method of collecting electronic tag information through a tag reader to generate a real-time odometer of the train includes: Start / wait for the tag reader to determine whether the tag information is read in real time; If yes, query the basic data to obtain real-time mileage information; If not, start / wait for the tag reader to form an electronic tag basic database, obtain real-time mileage information, form a real-time mileage information table, perform real-time corrections based on the real-time mileage information, and continue to determine whether the tag information is read in real time.

3. The method for accurately locating long and short chains based on image recognition according to claim 1, characterized in that: The method of obtaining the TLL signal count by the photoelectric encoder to generate the real-time odometer of the train includes: The TTL signal count is obtained from the photoelectric encoder and given to the camera component through the signal multiplication TTL signal. The mileage information is currently accumulated through the TTL signal to obtain the real-time mileage positioning information and form a real-time line information basic database. The real-time line information basic database is added to form a real-time mileage information table. Real-time correction is performed through the real-time mileage information table to generate a real-time tag positioning difference mileage table. The real-time mileage positioning information is obtained synchronously and the real-time line information basic database is updated. The electronic tag long and short chain is formed through the average value of the tag long and short chain and the correction difference of the electronic tag basic database.

4. The method for accurately locating long and short chains based on image recognition according to claim 1, characterized in that: The real-time calculation of the 100-meter mark includes: Waiting / starting the calculation of hundred-meter mark recognition, calculating the first mileage of the hundred-meter mark through sampling data, obtaining the real-time second mileage of the hundred-meter mark through the photoelectric encoder, obtaining the hundred-meter mark unique code generated by the unique identifier through the average value of the first mileage of the hundred-meter mark and the second mileage of the hundred-meter mark, and comparing the hundred-meter mark difference with the hundred-meter mark basic database, and judging whether the minimum preset difference is obtained; If so, automatically record the current section 100-meter mark length and short chain; If not, compare the average value of the first mileage of the hundred-meter mark and the second mileage of the hundred-meter mark with the hundred-meter mark difference in the basic database, wait / start the hundred-meter mark recognition calculation, and form the hundred-meter mark long and short chain through the average value of the label long and short chain and the corrected difference in the hundred-meter mark basic database.

5. The method for accurately locating long and short chains based on image recognition according to claim 1, characterized in that: The electronic tag real-time calculation includes: Waiting for / starting electronic tag identification calculation, calculating the first mileage of the electronic tag through sampling data, obtaining the real-time second mileage of the electronic tag through the photoelectric encoder, automatically generating the electronic tag unique code by comparing the average value of the first mileage of the electronic tag and the second mileage of the electronic tag with the electronic tag difference in the electronic tag basic database, and judging whether the minimum difference is found; If yes, automatically record the length of the electronic tag chain in the current section; If not, compare the average of the first mileage of the electronic tag and the second mileage of the electronic tag with the electronic tag difference in the electronic tag basic database, wait / start the electronic tag identification calculation, and form the electronic tag long and short chain through the average of the tag long and short chain in the electronic tag basic database and the corrected difference.

6. A long and short chain precise positioning method based on image recognition according to claim 4 or 5, characterized in that: The calculation of the difference between the length and short chain of each hundred-meter mark and / or electronic tag, and the generation of a statistical list of the length and short chain of the operation line segment according to the collection results of the first end point and the second end point of the train passing through, include: Correct the current 100-meter mark basic database through the 100-meter mark long and short chain, and automatically add the current 100-meter mark long and short chain to the dynamic long and short chain basic database; The electronic tag long-short chain formed by the average value of the electronic tag long-short chain and the difference value amends the current electronic tag basic database, and automatically adds the current electronic tag long-short chain to the dynamic long-short chain basic database; Determine whether the train starts from the first endpoint and arrives at the second endpoint. If so, end; if not, correct the current basic database through the long and short chains of the hundred-meter mark, calculate the long and short chain difference between each hundred-meter mark and / or electronic tag, and generate an automatic statistical list of long and short chains.

7. The method for accurately locating long and short chains based on image recognition according to claim 1, characterized in that: The electronic tags are installed at 100-meter points throughout the entire kilometer of the steel track bed to record and mark the position of each 100-meter mark; the photoelectric encoder is installed on the axle and also includes a linear array camera, which is installed in the middle of the train body and on both sides of the front of the train.

8. A long and short chain precise positioning system based on image recognition, used to execute the long and short chain precise positioning method based on image recognition as described in any one of claims 1 to 7, characterized in that: include: A hardware device system, wherein the hardware device system includes an electronic tag, a photoelectric encoder, and a high-definition 4K linear array camera; The software system includes a real-time data acquisition system and a real-time data analysis system; the real-time data acquisition system is used to collect real-time mileage data of the photoelectric encoder, and use the reader to capture the installed electronic tag information; and drive the linear array camera to capture the route label and hundred-meter mark information to build a complete mileage data route information library.

Images