Train operation path accurate identification method and system

Abstract

The application provides a train operation path accurate identification method and system, belongs to the path identification technical field, obtains an image to be identified; detects a turnout in the image, extracts a key turnout according to the turnout coordinate information; respectively carries out segmentation on the full scene of the image and the extracted key turnout, obtains a full scene image and a turnout image, splices the segmentation results according to the turnout pixel coordinate position, and uses the spliced result as a train multi-path identification result; according to the relative position of the vehicle-mounted camera, refers to the track pixel area close to the front end of the train in the image, determines a starting growth pixel point, uses a multi-neighbor region growth algorithm according to the starting pixel growth point, processes the train multi-path identification result, extracts the pixel point of the train operation path, and accurately identifies the final train operation path. The application realizes the identification of multiple paths, guarantees the high similarity of the pixel information of the same type of object, uses the region growth algorithm to process the train multi-path identification result, and realizes the accurate identification of the train path.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of path recognition, and in particular to a train running path accurate recognition method and system. BACKGROUND

[0002] In recent years, the frontiers of rail transit safety science research and deep learning technology are deeply integrated, and intelligent analysis methods based on computer vision have been widely applied in the field of rail transit safety monitoring. Among them, track detection as the basic link of train running environment perception has become the core research content of related intelligent algorithms. Although existing research has proposed various track feature extraction methods, most of them are limited to simple single-line railway scenes. In the face of complex line scenes containing switches, how to accurately extract track feature information, determine the opening direction of the switch, determine the train running path, and further identify the key area, these problems are still less researched and need to be further explored.

[0003] From the perspective of technical implementation path, the current track feature extraction method is mainly divided into two categories: one is based on traditional image processing method, and the other is based on deep learning semantic segmentation method. The traditional image processing method mainly uses the unique geometric features of the track to realize the track positioning through edge detection, Hough transform and other technologies. This kind of method has high calculation efficiency, but in the case of complex lighting conditions or the presence of interference, it is easy to be disturbed by the background environment and lead to detection failure. The semantic segmentation method based on deep learning can realize pixel-level track area extraction through end-to-end training, which significantly improves the detection accuracy, but in the complex scene of multiple parallel or intersecting tracks, it is difficult to accurately distinguish the target track from other tracks. At the same time, although the latest research on switch scenes can realize the classification and identification of switch direction through target detection technology, it generally lacks in-depth analysis of the actual train running path. Accurate identification of train running path not only relates to the real-time judgment of train driving direction, but also is an important basis for realizing the accurate positioning of forward obstacles, which has great practical significance for improving train running safety level and operation efficiency. Therefore, it is urgent to develop intelligent track detection and path recognition algorithm suitable for complex scenes. SUMMARY

[0004] The purpose of the present application is to provide a train running path accurate recognition method and system, which analyzes the prior geometric features of each component of the track, combines deep learning and traditional image processing method, and realizes the accurate recognition of the train running path, so as to solve at least one technical problem existing in the background technology.

[0005] In order to achieve the above purpose, the present application adopts the following technical scheme:

[0006] In the first aspect, the present application provides a train running path accurate recognition method, comprising:

[0007] acquire an image to be recognized;

[0008] processing the acquired image by using a pre-trained recognition model to obtain a final running path; wherein the recognition model comprises an extraction network, a segmentation network and a growth network; the extraction network is used for detecting a turnout in the image, and extracting a key turnout according to turnout coordinate information; the segmentation network is used for respectively segmenting a full-scene image and the extracted key turnout, obtaining a full-scene image and a turnout image, and splicing the segmentation results according to the pixel coordinate positions of the turnout to obtain a train multi-path recognition result; the growth network is used for determining a starting growth pixel point according to the relative position of a vehicle-mounted camera, referring to a track pixel area close to the front end of the train in the image, processing the train multi-path recognition result by using a multi-neighborhood region growth algorithm, extracting pixel points of the train running path, and accurately recognizing the final train running path.

[0009] As a further limitation of the first aspect of the application, training the extraction network comprises: acquiring multiple complex rail transit line images, labeling the turnout targets in the images, and constructing a turnout image target detection data set for deep learning training; training a target detection algorithm using the turnout image target detection data set to learn the image features of the turnout.

[0010] As a further limitation of the first aspect of the application, the training parameters and target detection data set in the target detection model are configured, the model training instruction is run, and the model is started to be trained, so that the model can complete target positioning and target classification.

[0011] As a further limitation of the first aspect of the application, the pixel position information of the left upper corner and the right lower corner of the coordinate frame is determined according to the detected turnout target frame; and the key turnout in the full-scene image is extracted according to the key pixel coordinate information of the turnout coordinate frame.

[0012] As a further limitation of the first aspect of the application, training the segmentation network comprises: labeling each component of the track in the image, and making a full-scene semantic segmentation data set; according to the coordinate information of the turnout, the original image and the labeled image in the full-scene semantic segmentation data set are respectively cropped to construct a turnout semantic segmentation data set; and the semantic segmentation model is trained using the full-scene semantic segmentation data set and the turnout semantic segmentation data set respectively.

[0013] As a further limitation of the first aspect of the application, the model training parameters in the semantic segmentation model and the corresponding semantic segmentation data set are configured respectively, the model training instruction is run, and the model is started to be trained, so that the model can effectively recognize the pixel information of each component of the track in the image.

[0014] Secondly, the application provides a train running path accurate recognition system, comprising:

[0015] An acquisition module is configured to acquire an image to be recognized.

[0016] A processing module is configured to process the acquired image by using a pre-trained recognition model to obtain a final running path, wherein the recognition model comprises an extraction network, a segmentation network and a growth network; the extraction network is configured to detect a turnout in the image and extract a key turnout according to turnout coordinate information; the segmentation network is configured to segment a full scene of the image and the extracted key turnout respectively to obtain a full scene image and a turnout image, and splice the segmentation results according to the pixel coordinate positions of the turnout to obtain a train multi-path recognition result; and the growth network is configured to determine a starting growth pixel point according to the relative position of a vehicle-mounted camera and refer to a track pixel area close to the front end of the train in the image, use a multi-neighborhood region growth algorithm to process the train multi-path recognition result, extract pixel points of the train running path and accurately recognize the final train running path.

[0017] In a third aspect, the present application provides a non-transitory computer readable storage medium for storing computer instructions, which, when executed by a processor, implement the train running path accurate recognition method according to the first aspect.

[0018] In a fourth aspect, the present application provides a computer device comprising a memory and a processor, wherein the processor and the memory are in communication with each other, the memory stores program instructions executable by the processor, and the processor invokes the program instructions to execute the train running path accurate recognition method according to the first aspect.

[0019] In a fifth aspect, the present application provides an electronic device comprising a processor, a memory and a computer program, wherein the processor is connected with the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to execute instructions for implementing the train running path accurate recognition method according to the first aspect.

[0020] The application has the advantages that: for the train running path recognition problem under a complex rail transit line, the changes of the characteristics of each component of the track under different train paths are fully tapped. First, the target detection and semantic segmentation method is used to realize the multi-path recognition, and the result guarantees the high similarity of the pixel information of the same category object. Further, according to the position super parameter of the vehicle-mounted camera, the pixel growth starting point is determined, the region growing algorithm is used to process the train multi-path recognition result, so that all the pixel points of the train running path area are extracted, and the accurate recognition of the train path is realized. The method does not depend on complex logical judgment, and can realize the recognition of the train running path through the spatial connectivity of the track area in the train running path. The method principle is simple, and with the reliable prior geometric characteristics, the functions and characteristics of the machine vision and the region growing algorithm can be combined to realize the accurate recognition of the train running path under the complex rail transit line, and the result is accurate and reliable.

[0021] The advantages of the additional aspects of the application will be more apparent from the following description part or be understood through the practice of the application. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the application, the drawings needed in the embodiment description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0023] Figure 1 The train running path recognition method flowchart described in the embodiment of the application.

[0024] Figure 2 The complex rail transit line data set diagram described in the embodiment of the application.

[0025] Figure 3 The train running path accurate recognition result schematic diagram described in the embodiment of the application. Among them, Figure 3 (a) is the original input image, Figure 3 (b) is the train multi-path recognition result, Figure 3 (c) is the train running path accurate recognition result. DETAILED DESCRIPTION

[0026] The embodiments of the application will be described in detail below, and the examples of the embodiments are shown in the drawings, wherein the same or similar reference signs represent the same or similar elements or elements with the same or similar functions throughout. The embodiments described below through the drawings are exemplary and are only used to explain the application, and cannot be explained as the limitation of the application.

[0027] As used herein, the terms "have," "has," "have," "having," "include," "includes," "including," "indude," "indudes," "induding," or "indudeing" are used to indicate one or more of the stated features, elements, steps, or components or groups thereof; are used to indicate that a stated feature, element, step or component can be included in some embodiments but not in others; and are used to indicate that a stated feature, element, step or component can or can not be present in the embodiment being described, and that, therefore, examples can include embodiments that "do not include" or "that do not "include" certain features, elements, steps, or components.

[0028] It should also be understood that, unless clearly indicated otherwise, terminology deeded as taken from a patent or other juridical document is to be taken in context, and is to be interpreted as consistent with conventional usage in the field of the present application, and is not to be interpreted in an idealized or overly formal sense unless so expressly defined herein.

[0029] As will be understood by one of ordinary skill in the art, the singular forms "a," "an," "said," and "the" include plural referents unless the context clearly dictates otherwise.

[0030] In the description of the present application, the terms "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" and the like, mean that a particular feature, structure, material, or characteristic is included in at least one embodiment or example of the present application, and that the feature, structure, material or characteristic can be

[0031] In order to facilitate the understanding of the present application, the present application will be further explained with specific embodiments in conjunction with the accompanying drawings, and the specific embodiments do not constitute limitations to the embodiments of the present application.

[0032] It should be understood by those skilled in the art that the drawings are only schematic of the embodiments, and the components in the drawings are not necessarily essential for the implementation of the present application.

[0033] Example 1

[0034] In this embodiment 1, first provide a train operation path accurate identification system, comprising: an acquisition module for acquiring an image to be identified; a processing module for processing the acquired image using a pre-trained identification model to obtain the final operation path; wherein the identification model comprises an extraction network, a segmentation network and a growth network; the extraction network is used to detect the turnout in the image, and the key turnout is extracted according to the turnout coordinate information; the segmentation network is used to segment the full scene of the image and the extracted key turnout respectively, to obtain the full scene image and the turnout image, and the segmentation results are spliced according to the turnout pixel coordinate position as the train multi-path identification result; the growth network is used to determine the starting growth pixel point according to the relative position of the vehicle-mounted camera, refer to the track pixel area close to the front end of the train in the image, and extract the pixel points of the train operation path by using the multi-neighborhood region growing algorithm to process the train multi-path identification result, and accurately identify the final train operation path.

[0035] The specific flow steps of the region growing algorithm are: first, select an initial seed point, then check the neighborhood pixels by iteration, if the pixel satisfies the similarity criterion color distance ||C(p)-C(q)||2≤T (where C(p) and C(q) are RGB color vectors, and T is a pre-set threshold), then merge it into the growing region, until there is no new pixel to merge, and finally output the connected region.

[0036] In this embodiment, the above system is used to realize a train operation path accurate identification method, comprising: using the acquisition module to acquire an image to be identified; using the processing module to process the acquired image using a pre-trained identification model to obtain the final operation path; wherein the identification model comprises an extraction network, a segmentation network and a growth network; the extraction network is used to detect the turnout in the image, and the key turnout is extracted according to the turnout coordinate information; the segmentation network is used to segment the full scene of the image and the extracted key turnout respectively, to obtain the full scene image and the turnout image, and the segmentation results are spliced according to the turnout pixel coordinate position as the train multi-path identification result; the growth network is used to determine the starting growth pixel point according to the relative position of the vehicle-mounted camera, refer to the track pixel area close to the front end of the train in the image, and extract the pixel points of the train operation path by using the multi-neighborhood region growing algorithm to process the train multi-path identification result, and accurately identify the final train operation path.

[0037] The training of the extraction network comprises: acquiring multiple complex rail transit line images, labeling the turnout targets in the images, and constructing a turnout image target detection data set for deep learning training; training a target detection algorithm using the turnout image target detection data set to learn the image features of the turnout. The training parameters of the target detection model and the target detection data set are configured, the model training instruction is run, and the model is started to train, so that the model can complete target positioning and target classification.

[0038] The pixel position information of the left upper corner and the right lower corner of the coordinate frame is determined according to the detected turnout target frame, and the key turnout in the full-scene image is extracted according to the key pixel coordinate information of the turnout coordinate frame.

[0039] The training of the segmentation network comprises: labeling each component of the track in the image, and making a full-scene semantic segmentation data set; according to the coordinate information of the turnout, the original image and the labeled image in the full-scene semantic segmentation data set are respectively cropped to construct a turnout semantic segmentation data set; the semantic segmentation model is trained using the full-scene semantic segmentation data set and the turnout semantic segmentation data set respectively. The model training parameters in the semantic segmentation model and the corresponding semantic segmentation data set are configured, the model training instruction is run, and the model is started to train, so that the model can effectively identify the pixel information of each component of the track in the image.

[0040] Embodiment 2

[0041] As shown in Figures 1 to 3 The embodiment 2 provides a train running path accurate identification method, which can realize the identification of the train running path under the complex rail transit line by virtue of reliable prior geometric features and the functional characteristics of machine vision and region growing algorithm, and the result is accurate and reliable.

[0042] The train running path accurate identification method provided by the embodiment comprises the following steps:

[0043] Step S1, using a target detection algorithm to detect the turnout existing in the image, and returning the turnout coordinate information, and extracting the key turnout in the original image according to the coordinate information.

[0044] The step S1 specifically comprises the following steps:

[0045] Step 1.1, collecting the complex rail transit line images collected by the front-end camera of the train, manually labeling the turnout targets in the images, and constructing a turnout image target detection data set for deep learning training.

[0046] Step 1.2, training an advanced target detection algorithm using the turnout image target detection data set, and learning the image features of the turnout.

[0047] Step 1.3, using the trained algorithm to detect the turnout existing in the image, according to the detected turnout target box, return the pixel position information (x min ,y min ,x max ,y max ) of the left upper corner and the right lower corner of the coordinate box.

[0048] Step 1.4, according to the key pixel coordinate information of the turnout coordinate box, extract the key turnout in the full scene image, which is used for further fine segmentation in the subsequent.

[0049] Step S2: using semantic segmentation algorithm to segment the full scene and the extracted key turnout respectively, and splicing the segmentation results according to the turnout pixel coordinate position as the train multi-path recognition result.

[0050] The step S2 specifically includes the following steps:

[0051] Step 2.1, according to the collected vehicle-mounted image, manually label each component of the track to make a full scene semantic segmentation dataset. According to the coordinate information of the turnout, the original image and the labeled image in the full scene semantic segmentation dataset are respectively cropped to construct the turnout semantic segmentation dataset.

[0052] Step 2.2, respectively using the full scene semantic segmentation dataset and the turnout semantic segmentation dataset to train an advanced semantic segmentation model;

[0053] Step 2.3, using the trained semantic segmentation model to infer and predict the full scene image and the turnout image respectively.

[0054] Step 2.4, according to the turnout pixel coordinate information, splicing the full scene segmentation result and the turnout segmentation result as the train multi-path recognition result.

[0055] Step S3: according to the position of the train front camera, determine the starting pixel growth coordinate point, use region growing method to process the train multi-path recognition result, and extract the specific running path of the train.

[0056] The step S3 specifically includes the following steps:

[0057] Step 3.1, according to the relative position of the vehicle-mounted camera, referring to the track pixel area close to the front end of the train in the image, determine the starting growth pixel point.

[0058] Step 3.2, according to the starting pixel growth point, using multi-neighbor region growing algorithm to process the train multi-path recognition result, extract the pixel point of the train running path, and accurately identify the train running path.

[0059] In summary, the embodiment is aimed at the problem of train running path recognition under complex rail transit lines. By relying on the prior geometric characteristics of the track, the embodiment innovatively combines deep learning algorithms and traditional image processing algorithms to achieve a precise train running path recognition method. The method analyzes the track structure characteristics of switches in different directions to accurately identify the geometric relationship of each part of the track, thereby determining the spatial connection characteristics between pixel points in the train running path. First, pixel classification of each component of the track is achieved using target detection and semantic segmentation. Then, referring to the position parameter of the front camera of the train, the region growing algorithm is used to extract all pixel points in the train running path, completing the accurate recognition of the train running path.

[0060] Embodiment 3

[0061] The embodiment 3 provides a non-transitory computer readable storage medium for storing computer instructions, which, when executed by a processor, implement the precise train running path recognition method as described above. The method includes:

[0062] Obtaining an image to be recognized;

[0063] Processing the obtained image using a pre-trained recognition model to obtain the final running path. The recognition model includes an extraction network, a segmentation network, and a growth network. The extraction network is used to detect switches in the image and extract key switches according to switch coordinate information. The segmentation network is used to segment the full scene and the extracted key switches in the image, obtaining a full scene image and a switch image. The segmentation results are spliced according to the switch pixel coordinate position as the train multi-path recognition result. The growth network is used to determine the starting growth pixel point according to the relative position of the on-board camera and the track pixel area near the front end of the train in the image. Based on the starting pixel growth point, the multi-neighbor region growing algorithm is used to process the train multi-path recognition result, extract the pixel points of the train running path, and accurately recognize the final train running path.

[0064] Embodiment 4

[0065] The embodiment 4 provides a computer device, including a memory and a processor, the processor and the memory communicate with each other, the memory stores program instructions executable by the processor, and the processor calls the program instructions to execute the precise train running path recognition method as described above. The method includes:

[0066] Obtaining an image to be recognized;

[0067] The obtained image is processed by using a pre-trained recognition model to obtain a final running path; wherein the recognition model comprises an extraction network, a segmentation network and a growth network; the extraction network is used for detecting a turnout in the image, and extracting a key turnout according to turnout coordinate information; the segmentation network is used for respectively segmenting a full scene of the image and the extracted key turnout to obtain a full scene image and a turnout image, and splicing the segmentation results according to the pixel coordinate positions of the turnout as a train multi-path recognition result; the growth network is used for determining a starting growth pixel point according to the relative position of the vehicle-mounted camera and referring to a track pixel area close to the front end of the train in the image, processing the train multi-path recognition result by using a multi-neighborhood region growth algorithm, extracting pixel points of the train running path, and accurately recognizing the final train running path.

[0068] Embodiment 5

[0069] The embodiment 5 provides an electronic device, comprising a processor, a memory and a computer program; wherein the processor is connected with the memory, and the computer program is stored in the memory; when the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device executes instructions for realizing a train running path accurate identification method as described above, the method comprising:

[0070] obtaining an image to be identified;

[0071] processing the obtained image by using a pre-trained recognition model to obtain a final running path; wherein the recognition model comprises an extraction network, a segmentation network and a growth network; the extraction network is used for detecting a turnout in the image, and extracting a key turnout according to turnout coordinate information; the segmentation network is used for respectively segmenting a full scene of the image and the extracted key turnout to obtain a full scene image and a turnout image, and splicing the segmentation results according to the pixel coordinate positions of the turnout as a train multi-path recognition result; the growth network is used for determining a starting growth pixel point according to the relative position of the vehicle-mounted camera and referring to a track pixel area close to the front end of the train in the image, processing the train multi-path recognition result by using a multi-neighborhood region growth algorithm, extracting pixel points of the train running path, and accurately recognizing the final train running path.

[0072] Those skilled in the art will understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0073] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flow or blocks Figure 1 means for functionally implementing the steps in one or more flow or blocks.

[0074] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart block or blocks. Figure 1 one or more flow or blocks Figure 1 means for functionally implementing the steps in one or more flow or blocks.

[0075] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flow or blocks Figure 1 means for functionally implementing the steps in one or more flow or blocks.

[0076] The above description is only a specific implementation of the present application, and is not intended to limit the protection scope of the present application. It should be understood by those skilled in the art that various modifications or changes can be made to the disclosed technical solutions without departing from the spirit and scope of the present application.

Claims

1. A train operation path accurate identification method, characterized in that, The method comprises the following steps: acquiring an image to be recognized; processing the acquired image by using a pre-trained recognition model to obtain a final running path; wherein the recognition model comprises an extraction network, a segmentation network and a growth network; the extraction network is used for detecting a turnout in the image and extracting a key turnout according to turnout coordinate information; the segmentation network is used for respectively segmenting a full-scene image and the extracted key turnout to obtain a full-scene image and a turnout image, and splicing the segmentation results according to the pixel coordinate positions of the turnout as a train multi-path recognition result; and the growth network is used for determining a starting growth pixel point according to the relative position of a vehicle-mounted camera and referring to a track pixel area close to the front end of the train in the image, processing the train multi-path recognition result by using a multi-neighborhood region growth algorithm, extracting pixel points of a train running path and accurately recognizing the final train running path.

2. The train operation path precise recognition method according to claim 1, characterized in that, The training of the extraction network comprises the following steps: acquiring multiple complex rail transit line images, labeling a turnout target in the images, and constructing a turnout image target detection data set for deep learning training; training a target detection algorithm by using the turnout image target detection data set to learn the image features of the turnout.

3. The train operation path precise recognition method according to claim 2, characterized in that, The training parameters and the target detection data set in the target detection model are configured, a model training instruction is run, and the model is started to be trained, so that the model completes target positioning and target classification.

4. The train operation path precise recognition method according to claim 1, characterized in that, According to the detected turnout target box, the pixel position information of the left upper corner and the right lower corner of the coordinate box is determined; and according to the key pixel coordinate information of the turnout coordinate box, a key turnout in the full-scene image is extracted.

5. The train operation path precise recognition method according to claim 1, characterized in that, The training of the segmentation network comprises the following steps: labeling each component of the track in the image to make a full-scene semantic segmentation data set; according to the coordinate information of the turnout, the original image and the labeled image in the full-scene semantic segmentation data set are respectively cropped to construct a turnout semantic segmentation data set; and a semantic segmentation model is trained by using the full-scene semantic segmentation data set and the turnout semantic segmentation data set respectively.

6. The train operation path precise recognition method according to claim 5, characterized in that, The model training parameters in the semantic segmentation model and the corresponding semantic segmentation data set are respectively configured, a model training instruction is run, and the model is started to be trained, so that the model recognizes the pixel information of each component of the track in the image.

7. A train operation path precision identification system, characterized in that, The method comprises the following steps: an acquisition module is configured to acquire an image to be recognized; a processing module is configured to process the acquired image by using a pre-trained recognition model to obtain a final running path; wherein the recognition model comprises an extraction network, a segmentation network and a growth network; the extraction network is used for detecting a turnout in the image and extracting a key turnout according to turnout coordinate information; the segmentation network is used for respectively segmenting a full-scene image and the extracted key turnout to obtain a full-scene image and a turnout image, and splicing the segmentation results according to the pixel coordinate positions of the turnout as a train multi-path recognition result; and the growth network is used for determining a starting growth pixel point according to the relative position of a vehicle-mounted camera and referring to a track pixel area close to the front end of the train in the image, processing the train multi-path recognition result by using a multi-neighborhood region growth algorithm, extracting pixel points of a train running path and accurately recognizing the final train running path.

8. A non-transitory computer-readable storage medium, comprising: The non-transitory computer readable storage medium is configured to store computer instructions, which, when executed by a processor, implement the train operation path accurate identification method according to any one of claims 1-6.

9. A computer device, comprising: A device includes a memory and a processor, the processor and the memory communicate with each other, the memory stores program instructions executable by the processor, and the processor invokes the program instructions to execute the train operation path accurate identification method according to any one of claims 1-6.

10. An electronic device, comprising: The device includes: A processor, a memory, and a computer program; wherein the processor is connected with the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to execute instructions for implementing the train operation path accurate identification method according to any one of claims 1-6.

Images