An image recognition-based automatic road sign counting method

Abstract

The application discloses a kind of road sign automatic statistics methods based on image recognition, comprising the following steps: obtaining the traffic engineering plan design drawing electronic paper of data to be counted;Traffic engineering plan design drawing electronic paper is input to road sign target detection model, and the sign feature is obtained;Traffic engineering plan design drawing electronic paper is input to road segmentation model, and the road feature is obtained;According to the sign feature data type, the sign name is obtained;Sign feature data coordinates and sign feature image are input into road sign character recognition model, and the road post number and support form are obtained;Sign feature data coordinates, road feature image, road feature data coordinates and road post number data coordinates are input into road sign position discrimination model, and the setting position is obtained, to complete the automatic statistics of all required information of road sign.The statistical method of the application effectively avoids the omissions that may occur in the manual operation process, thereby improving work efficiency and design accuracy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence and image recognition, and particularly relates to a road sign automatic statistical method based on image recognition. BACKGROUND

[0002] With the high-quality development of China's economy and society, the living standards of residents have improved significantly, and the number of motor vehicles has grown rapidly, which has put higher demands on urban transportation systems. In order to cope with the challenges brought by the growth of traffic volume, fine design of transportation engineering is particularly important, and how to effectively improve the design efficiency and drawing quality in the road construction drawing design stage has become a problem to be solved.

[0003] After the completion of transportation engineering design, the relevant information in the transportation engineering plan design drawing needs to be counted, including sign post number, sign name, setting position, support form, etc. At present, this process mainly adopts the traditional manual statistical method, and the designer needs to spend a lot of time checking and filling in the information of the sign item by item when drawing, which not only is inefficient, but also is prone to omissions due to human factors, and is difficult to meet the current demand for efficient and high-quality design.

[0004] The prior art lacks practical tools and effective methods for automatic processing in this work, and it is difficult to cope with the growing workload of road design. Therefore, how to develop an efficient and accurate road sign statistical method to replace the traditional manual statistical method and improve the design efficiency and drawing quality has become one of the key problems to be solved in road construction drawing design technology. This has important practical significance for improving road design quality, shortening design cycle and better serving the construction of urban transportation systems. SUMMARY

[0005] The purpose of the present application is to overcome the shortcomings of the background art, and to provide a road sign automatic statistical method based on image recognition, which is efficient and accurate, and can replace the traditional manual statistical method to improve the design efficiency and accuracy.

[0006] To achieve the above purpose, the road sign automatic statistical method based on image recognition provided by the present application comprises the following steps:

[0007] Obtain the electronic drawing of the transportation engineering plan design drawing to be counted;

[0008] Input the electronic drawing of the transportation engineering plan design drawing into the trained road sign target detection model to obtain the sign features; the sign features include sign feature images, sign feature data types and sign feature data coordinates;

[0009] The traffic engineering plan design drawing electronic paper is input into the trained road segmentation model to obtain road features; the road features include road feature images, road feature data coordinates, and road stake data coordinates;

[0010] A sign name is obtained according to the sign feature data type; the sign feature data coordinates and the sign feature images are input into a road sign character recognition model to obtain road stakes and support forms;

[0011] The sign feature data coordinates, the road feature images, the road feature data coordinates, and the road stake data coordinates are input into a road sign position discrimination model to obtain a setting position, so as to complete automatic statistics of all required information of the road sign.

[0012] As a preferred embodiment, the traffic engineering plan design drawing to be counted contains design information of road routes and road signs.

[0013] As a preferred embodiment, the traffic engineering plan design drawing to be counted adopts an image format.

[0014] As a preferred embodiment, the traffic engineering plan design drawing to be counted is obtained by performing data cleaning on the traffic engineering plan design drawing and deleting irrelevant graphic and text data.

[0015] As a preferred embodiment, the road sign target detection model is trained and established in the following manner:

[0016] The road sign target detection model is obtained by training a target detection model using a road sign data set;

[0017] The road sign data set is a target detection data set, and each data is labeled with four vertices.

[0018] The labeling range of the road sign data set includes image information of sign names, text information of stake numbers, and text information of support forms.

[0019] The road sign data set is labeled using “Road Traffic Signs and Markings Part 2: Road Traffic Signs” (GB5768.2-2022) and traffic engineering plan design drawings of multiple engineering projects, including 48 types of prohibition signs, 43 types of indication signs, 47 types of warning signs, 77 types of direction signs, 17 types of tourism signs, 10 types of notice signs, and 20 types of auxiliary signs.

[0020] As a preferred embodiment, the road segmentation model is trained and established in the following manner:

[0021] The road segmentation model is obtained by training a semantic segmentation model using a road route segmentation data set.

[0022] The road route segmentation dataset is a semantic segmentation dataset.

[0023] The annotation range of the road route segmentation dataset is a center line and a road stake number.

[0024] As a preferred embodiment, the road sign text recognition model is trained and established by the following method:

[0025] The road sign text recognition model is composed of a text recognition model, a similarity text matching algorithm and a road sign text library.

[0026] After the text recognition model recognizes the sign name and support form text, the similarity text matching algorithm is used to match the closest text in the road sign text library.

[0027] The road sign text library includes two types of text, namely sign name and support form.

[0028] As a preferred embodiment, the road sign position discrimination model is a vector cross product calculation model.

[0029] As a preferred embodiment, the road sign position discrimination model is trained and established by the following method:

[0030] The road sign position discrimination model first samples four directed line segments close to the position to be judged from the road feature data.

[0031] The positive direction of the directed line segment is determined according to the road stake number, the starting point is a point near the smaller road stake number, the ending point is a point near the larger road stake number, and the positive direction is from the starting point to the ending point.

[0032] Then, the vector cross product of the four vertices of the road sign to be judged and the four directed line segments is calculated, and a total of 16 cross product results are obtained.

[0033] If more than half of the cross product results of the vertices are greater than 0, the road sign is on the left side of the road, otherwise it is on the right side of the road.

[0034] As a preferred embodiment, the road stake number is recognized from the road feature image and the road stake number data by using the text recognition model.

[0035] Compared with the prior art, the present application has the following advantages:

[0036] Firstly, the method of this invention is a road sign statistics method used in the road construction drawing design stage. It can automatically collect relevant information about road signs, including sign station number, sign name, installation location, support type, etc. This method greatly reduces the time required for designers to collect signs after traffic engineering design, replacing the traditional manual statistics method, effectively avoiding omissions that may occur during manual operation, thereby improving work efficiency and design accuracy. Through the application of this invention, designers can devote more energy to scheme optimization and design quality improvement, significantly shortening the design cycle and meeting the needs of efficient and high-quality design.

[0037] Secondly, this invention provides an efficient and automated method for statistical analysis of road signs, filling a gap in existing technologies for road construction drawing design. It plays a significant role in improving the quality of urban road design, optimizing the construction drawing design process, and promoting the construction of urban transportation systems. Attached Figure Description

[0038] Figure 1 This is a flowchart illustrating an automated statistical method for road signs based on image recognition.

[0039] Figure 2 This is a schematic diagram of a marker feature extraction method;

[0040] Figure 3 This is a schematic diagram of a road feature extraction method. Detailed Implementation

[0041] The following examples illustrate the implementation of the present invention in detail, but they do not constitute a limitation on the invention and are merely illustrative. Furthermore, the advantages of the present invention will become clearer and easier to understand by explaining them.

[0042] like Figure 1 As shown, after traffic engineering design, it is necessary to collect information such as sign station numbers, sign names, locations, and support types from the traffic engineering plan. Currently, this is done manually, requiring designers to spend a significant amount of time on this task during the drawing process, resulting in low efficiency. Therefore, this invention, based on thorough research and practical engineering experience, develops an efficient and accurate automated sign collection method to replace the traditional manual method, improving design efficiency and accuracy.

[0043] The present invention provides an automated road sign counting method based on image recognition, comprising the following steps:

[0044] S1: Obtain the traffic engineering plan design drawings to be used for statistical data;

[0045] Among them, the schematic diagram of the traffic engineering plan design to be statistically analyzed is as follows: Figure 2, Figure 3 The traffic engineering plan design drawings are shown in the image. The traffic engineering plan design drawings to be statistically analyzed are in image form. The method of obtaining the traffic engineering plan design drawings to be statistically analyzed is to clean the data of the traffic engineering plan design drawings, delete some irrelevant graphic and textual data information, and retain only key design information such as road routes and road signs (including corresponding markers).

[0046] S2: Input the traffic engineering plan design drawing into the trained road sign target detection model to obtain sign features. Input the traffic engineering plan design drawing into the trained road segmentation model to obtain road features;

[0047] Among them, such as Figure 2 As shown, the marker features include a marker feature image, a marker feature data type, and marker feature data coordinates. For example... Figure 3 As shown, the road features include road feature images and road feature data coordinates.

[0048] In this embodiment, the trained road sign object detection model is a YOLOv8 model (You Only Look Once version 8). The road sign dataset is annotated using the "Road Traffic Signs and Markings Part 2: Road Traffic Signs" (GB5768.2-2022) and traffic engineering plan design drawings from multiple engineering projects. The road sign dataset is an object detection dataset, with each data point labeled with four vertices. The categories of the road sign dataset include: 48 categories of prohibitory signs, 43 categories of instruction signs, 47 categories of warning signs, 77 categories of guide signs, 17 categories of tourist signs, 10 categories of notice signs, and 20 categories of auxiliary signs. Before training, the annotated drawings are cropped to meet the model's input size. The drawings are cropped into 512*512 pixel blocks.

[0049] In this embodiment, the trained road segmentation model uses the Unet model; the dataset consists of annotated traffic engineering plan design drawings from multiple engineering projects. Before training, the annotated drawings are cropped to meet the model's input dimensions. The drawings are cropped into 512*512 pixel blocks.

[0050] In this embodiment, the object detection model uses YOLOv8 and the semantic segmentation model uses Unet, but other models and data can be used in other implementations.

[0051] S3: Obtain the sign name based on the sign feature data type; input the sign feature data coordinates and the sign feature image into the road sign text recognition model to obtain the road station number and support type;

[0052] The sign text recognition model in this embodiment is composed of a text recognition model, a similarity text matching algorithm, and a road sign text library.

[0053] The text recognition model in this embodiment is an easyOCR model, and the similarity text matching algorithm in this embodiment adopts a cosine similarity algorithm using the vector output by bert-base-chinese. The road sign text library in this embodiment is established according to the text information counted in the sign list of multiple engineering projects.

[0054] The text recognition model in this embodiment is an easyOCR model, the similarity text matching algorithm adopts a cosine similarity algorithm using the vector output by bert-base-chinese, and the road sign text library is established according to the text information counted in the sign list of multiple engineering projects, but other models and data can be used in other embodiments.

[0055] S4: inputting the sign feature data coordinates, the road feature image, the road feature data coordinates, and the road stake number data coordinates into a road sign position discrimination model to obtain the sign position, thereby completing automatic counting of all required information of the road sign;

[0056] The principle of the road sign position discrimination model in this embodiment is as follows:

[0057] First, define the vertices of the rectangle, assuming that the four vertices of the rectangle are P1, P2, P3, and P4. Then, sample four line segments close to the position to be judged from the road feature data, and define the two endpoints A1 and B1, A2 and B2, A3 and B3, and A4 and B4 of each line segment. Take a line segment as an example, assuming that the two endpoints of the line segment are A(ax, ay) and B(bx, by). Then, calculate the direction of each rectangle vertex relative to the line segment: for each rectangle vertex P(px, py), calculate the cross product of vectors AB and AP. The cross product formula is: AB x AP = (bx-ax)*(py-ay)-(by-ay)*(px-ax). Then, judge the result of the cross product. If more than 8 of the 16 vertices have a cross product result greater than 0, it means that all the vertices of the rectangle are on the left side of the line segment. Otherwise, it means that all the vertices of the rectangle are on the right side of the line segment.

[0058] The text recognition model in this embodiment is an easyOCR model. Note that the text recognition model in this embodiment is an easyOCR model, but other models and data can be used in other embodiments.

[0059] The automatically counted information is shown in Table 1 as a typical case:

[0060] Table 1: Typical case of automatically counted information

[0061]

[0062] The above detailed description of the embodiments of the present application is made in conjunction with the accompanying drawings, but the present application is not limited to the above-described embodiments, and various changes can be made within the knowledge of those skilled in the art without departing from the spirit of the present application, and should be considered as falling within the scope of the present application.

[0063] The above is only a specific embodiment of the present application, and it should be noted that any changes or replacements that can be easily thought of by those skilled in the art within the technical scope disclosed by the present application should be covered within the protection scope of the present application, and the rest that is not described in detail belongs to the prior art.

Claims

1. An image recognition-based automatic statistical method for road signs, characterized in that: The method comprises the following steps: obtaining an electronic drawing of a traffic engineering plan to be counted; inputting the electronic drawing of the traffic engineering plan into a trained road sign target detection model to obtain sign features; the sign features include sign feature images, sign feature data types, and sign feature data coordinates; inputting the electronic drawing of the traffic engineering plan into a trained road segmentation model to obtain road features; the road features include road feature images, road feature data coordinates, and road stake data coordinates; obtaining a sign name according to the sign feature data type; inputting the sign feature data coordinates and the sign feature images into a road sign character recognition model to obtain a road stake number and a support form; inputting the sign feature data coordinates, the road feature images, the road feature data coordinates, and the road stake data coordinates into a road sign position discrimination model to obtain a setting position, thereby completing automatic counting of all required information of the road sign.

2. The image recognition-based road sign automation counting method according to claim 1, characterized in that: The traffic engineering plan to be counted contains design information of road routes and road signs.

3. The image recognition based road sign automation statistical method according to claim 2, characterized in that: The traffic engineering plan to be counted adopts an image format.

4. The image recognition based road sign automation statistical method according to claim 3, characterized in that: The traffic engineering plan to be counted is obtained by data cleaning of a traffic engineering plan.

5. The image recognition based automatic road sign statistics method according to any one of claims 1 to 4, characterized in that: The road sign target detection model is trained and established by the following method: The road sign target detection model is trained and obtained by a target detection model using a road sign data set; The road sign data set is a target detection data set, and each data is labeled with four vertices; The labeling range of the road sign data set includes image information of a sign name, text information of a stake number, and text information of a support form.

6. The image recognition based road sign automation statistical method according to claim 5, characterized in that: The road segmentation model is trained and established by the following method: The road segmentation model is trained and obtained by a semantic segmentation model using a road route segmentation data set; The road route segmentation data set is a semantic segmentation data set; The labeling range of the road route segmentation data set is a center line and a road stake number.

7. The image recognition based road sign automation statistical method according to claim 6, characterized in that: The road sign character recognition model is trained and established by the following method: The road sign character recognition model is composed of a character recognition model, a similarity text matching algorithm, and a road sign text library; After the character recognition model recognizes a sign name and a support form text, the similarity text matching algorithm is used to match the closest text in the road sign text library; The road sign text library includes two types of texts, i.e., a sign name and a support form.

8. The image recognition based road sign automation statistical method according to claim 7, characterized in that: The road sign position discrimination model is a vector cross product calculation model.

9. The image recognition based road sign automation statistical method according to claim 8, characterized in that: The road sign position discrimination model is trained and established by the following method: The road sign position discrimination model first samples four directed line segments close to a position to be judged from road feature data; The positive direction of the directed line segment is determined according to a road stake number; the starting point is a point near a smaller road stake number, the ending point is a point near a larger road stake number, and the positive direction is from the starting point to the ending point; Subsequently, the vector cross product of four vertices of a road sign at the position to be judged and the four directed line segments is calculated. If the dot product result of more than half of the vertices are greater than 0, the road sign is on the left side of the road, otherwise on the right side of the road.

10. The image recognition based road sign automation statistical method according to claim 9, characterized in that: The road post number digits are recognized from the road feature image and the road post number data by using a character recognition model.

Images