A road disease intelligent identification and early warning system

Abstract

The present invention discloses an intelligent road disease identification and early warning system, which relates to the field of road inspection technology and solves the problem that in the process of image feature extraction, the pixel feature analysis of road diseases under complex backgrounds is not accurate enough, which easily leads to misjudgment or omission of abnormal areas. The present invention uses a clustering algorithm to accurately screen out abnormal pixel points through orderly analysis of pixel points in the partitioned area. Compared with traditional methods, the ability to capture subtle pixel features of diseases is effectively improved; the image different-area calibration end and the feature verification processing end further integrate the abnormal pixel point information, and determine the feature value of the abnormal area in combination with the spatial position feature, which can accurately lock the spatial range and height difference of the disease area, providing a strong guarantee for the accurate identification of the disease, and according to the specific spatial position identified, the spatial position point associated with the corresponding image is locked, and the specific position feature of the corresponding abnormal area is quickly locked, thereby achieving more accurate feature recognition results.

Description

Technical Field

[0001] The present invention relates to the technical field of road inspection, and in particular to an intelligent road hazard identification and early warning system. Background Art

[0002] With the rapid development of transportation infrastructure, roads are playing an increasingly important role in economic and social operations. However, due to the long-term impact of factors such as vehicle loads and environmental erosion, road damage is becoming increasingly prominent. Traditional road damage detection relies primarily on manual inspections. This method is not only inefficient, but also subject to subjective judgment by inspectors and difficult to detect subtle damage in its early stages. Furthermore, manual inspections pose certain safety risks and cannot meet the requirements of modern transportation for timely and accurate road maintenance.

[0003] The application with publication number CN119808654A discloses a road and bridge disease analysis and early warning method and system based on big data, which relates to the field of smart city technology. By deploying multiple groups of sensors, the collected relevant lateral wind load information data is associated with the relevant vertical traffic load information data, and the external load influence coefficient Xhz is constructed. After comparison, a preliminary early warning signal is issued, and the relevant surface state data information and relevant vibration state data information of the current road and bridge are further collected and associated with the external load influence coefficient Xhz. Combined with the disease assessment model, the comprehensive disease degree index Zbh is calculated and compared with the assessment threshold K to determine whether there is a disease risk for the road and bridge, so as to generate and execute the corresponding level of disease warning instructions. Through the dynamic association and intelligent analysis of multi-dimensional data, accurate identification and real-time early warning of road and bridge diseases are achieved.

[0004] Some existing road defect recognition technologies use image processing methods. Although this has improved detection efficiency to a certain extent, it still has many shortcomings. For example, during the image feature extraction process, the pixel feature analysis of road defects in complex backgrounds is not accurate enough, which can easily lead to misjudgment or omission of abnormal areas. When determining the spatial location and characteristic values ​​of defects, there is a lack of systematic and scientific calculation models, making it difficult to accurately quantify the scale and severity of the defects.

[0005] In addition, most existing technologies lack a complete integrated solution from image acquisition to disease warning display, and are unable to provide comprehensive and real-time decision-making support for road management departments, resulting in irrational allocation of road maintenance resources and untimely disease repair, affecting road service life and traffic safety.

[0006] Therefore, there is an urgent need for an efficient, accurate and intelligent road hazard identification and early warning system to meet the needs of modern road maintenance and management. Summary of the Invention

[0007] In response to the shortcomings of the existing technology, the present invention provides an intelligent road disease identification and early warning system, which solves the problem that in the process of image feature extraction, the pixel feature analysis of road diseases in complex backgrounds is not accurate enough, which easily leads to misjudgment or omission of abnormal areas.

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions: a road hazard intelligent identification and early warning system, comprising:

[0009] The image feature processing end grayscales the corresponding image based on the acquired image of the relevant area, confirms the grayscale image, and then calibrates the abnormal pixels based on the pixel value features associated with different points in the grayscale image. The specific method is as follows:

[0010] Grayscale processing is performed on the acquired image of the relevant area to confirm the grayscale image, and then the confirmed grayscale image is divided into a plurality of different partitioned areas. When the area is divided, a preset equalization template is provided. The equalization template contains a plurality of equalization grids of the same area, and each equalization grid corresponds to a group of partitioned areas.

[0011] Determine the center point of the partitioned area and record it as the reference point. Then, determine the feature point farthest from the reference point in the partitioned area. Determine the feature line between the reference point and the feature point. Record the pixels on the feature line as the line pixel points. Arrange the line pixel points in order from near to far from the reference point to determine a set of line pixel point arrangement.

[0012] Then, the edge pixels in the partitioned area are processed in the same manner to determine the different line pixel arrangement sets associated with different edge pixels;

[0013] Pixel points at the same sorting position are selected from a set of connected pixel points, pixel values ​​associated with the corresponding pixel points at the same sorting position are confirmed, and feature verification is performed on the confirmed pixel values: a group of pixel values ​​is randomly selected as the intermediate pixel value Zh, and the clustering interval associated with the pixel value is confirmed using: [Zh-Y1, Zh+Y1], where Y1 is a preset value, and the pixel value belonging to this clustering interval is used as the clustering pixel value of this pixel value, different pixel values ​​are selected in turn as the intermediate pixel values, and the clustering pixel values ​​are gradually determined, and a group of clustering intervals with the largest total number of clustering pixel values ​​is selected as the standard interval, and based on the confirmed standard interval, the pixel values ​​that do not belong to this standard interval are calibrated as abnormal pixel values, and the pixel points associated with the abnormal pixel values ​​are calibrated as abnormal pixel points;

[0014] and calibrating abnormal pixels for several pixels at different sorting positions in several connected pixel arrangement sets;

[0015] The image different area calibration end calibrates the adjacent abnormal pixel points for the several abnormal pixel points calibrated in the image of the relevant area, and confirms that the specific area associated with the adjacent abnormal pixel points is recorded as the abnormal area;

[0016] The feature verification processing end determines the spatial position features associated with different pixels in the abnormal area based on several abnormal areas marked in the relevant area image. Based on the confirmed spatial position features, the feature value of the abnormal area is locked. The specific method is as follows:

[0017] Confirm the spatial position of the camera, confirm the vertical point of this spatial position from the confirmed relevant area image, record the pixel point associated with this vertical point as the feature check point, record the pixel value associated with the feature check point as Xd, and record the vertical distance associated with the feature check point as Jl;

[0018] Then the pixel values ​​associated with different pixel points in the abnormal area are calibrated as X k , where k represents different pixel points, using: |Xd-X k |×C1+Jl=J k Confirm the feature distance J associated with the corresponding pixel point k , where C1 is the preset fixed coefficient factor;

[0019] Confirm the horizontal distance between the corresponding pixel point and the feature verification point in the image of the relevant area. Based on the determined horizontal distance and the location of the corresponding pixel point, calibrate the pixel point in the same plane as the feature verification point. The calibrated position is the virtual position. Then, construct a set of specific vertical lines perpendicular to the virtual position and calibrate them as virtual vertical lines. Confirm the straight-line distance between different points and the spatial position of the shooting camera from the virtual vertical lines. Confirm the straight-line distance = J k The point is used as the spatial position feature point of this pixel;

[0020] Then confirm the spatial position feature points associated with different pixel points in the abnormal area in turn;

[0021] Based on the spatial position feature points associated with different pixel points in the same group of abnormal areas, a group of lowest points and a group of highest points are confirmed, and then the horizontal base plane where the lowest points are located is confirmed, and the vertical distance associated with the highest point and the horizontal base plane is locked. The locked vertical distance is used as the characteristic value of this abnormal area.

[0022] Preferably, it also includes:

[0023] The regional image acquisition end confirms the relevant regional images acquired by the image survey vehicle and transmits the confirmed relevant regional images to the image feature processing end.

[0024] Preferably, it also includes:

[0025] The area calibration display terminal confirms the disease of different abnormal areas based on the different characteristic values ​​associated with different abnormal areas, locks the diseased areas and displays them. The specific method is as follows:

[0026] The different feature values ​​associated with different abnormal areas in the relevant area image are calibrated as TZ q , where q represents different abnormal regions, will satisfy: TZ q Abnormal areas with a value ≥ Y2 are marked as diseased areas. Otherwise, no marking is performed and Y2 is the preset value.

[0027] The present invention provides an intelligent road hazard identification and early warning system. Compared with the existing technology, it has the following advantages:

[0028] This method uses a clustering algorithm to accurately screen out abnormal pixels through an orderly analysis of pixels within a partitioned area. Compared with traditional methods, it effectively improves the ability to capture subtle pixel features of diseased areas. The image calibration end and feature verification processing end further integrate abnormal pixel information and determine the characteristic value of the abnormal area based on spatial location characteristics. This can accurately locate the spatial range and height difference of the diseased area, providing a strong guarantee for accurate disease identification.

[0029] Based on the specific image associated with the corresponding road, the pixel features associated with the relevant pixel points of the corresponding image are confirmed, and then based on the pixel features associated with the corresponding points in the abnormal area, feature verification is performed to identify the spatial position of the corresponding point. Then, based on the identified specific spatial position, the spatial position point associated with the corresponding image is locked. The specific position features of the corresponding abnormal area can be quickly locked, thereby achieving more accurate feature recognition results, making the diseased area more accurately identified and achieving better recognition and early warning effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a schematic diagram of the principle framework of the present invention;

[0031] Figure 2 It is a schematic diagram showing the partition areas of the present invention. DETAILED DESCRIPTION

[0032] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0033] First embodiment

[0034] See also Figure 1 The present application provides a road disease intelligent identification and early warning system, including a regional image acquisition end, an image feature processing end, an image cross-region calibration end, a feature verification processing end, and a regional calibration display end, wherein the regional image acquisition end, the image feature processing end, the image cross-region calibration end, and the feature verification processing end are electrically connected from the output node to the input node in sequence, the eggplant feature verification processing end is electrically connected to the input node of the regional calibration display end, and the image cross-region calibration end is electrically connected to the input node of the regional calibration display end;

[0035] Among them, the regional image acquisition end confirms the relevant regional images acquired by the image survey vehicle and transmits the confirmed relevant regional images to the image feature processing end. Specifically, when performing road disease identification, the relevant images of the road are obtained and confirmed based on the corresponding vehicle. Then, the image features are verified from the confirmed road-related images, and abnormal points are identified from the corresponding regional images and specifically calibrated.

[0036] Among them, the image feature processing end, based on the acquired image of the relevant area, grayscales the corresponding image, confirms the grayscale image, and then calibrates the existing abnormal pixels based on the pixel value features associated with different points in the grayscale image. The specific method of calibration is as follows:

[0037] The acquired images of the relevant areas are gray-scaled to confirm the gray-scaled images (the gray-scaled images are gray-scaled here to facilitate subsequent feature confirmation). The confirmed gray-scaled images are then divided into equal areas to confirm several different partitioned areas. A preset equalization template exists for area equalization. The equalization template contains several equalization grids of the same area. Each equalization grid corresponds to a group of partitioned areas. The area of ​​each partitioned area is the same, and the partitioned areas are confirmed based on a specific template. When the corresponding camera acquires the image, there is a corresponding acquisition angle, and the height of the vehicle is unchanged. Therefore, the size of the image acquired each time is consistent. Therefore, based on the corresponding equalization template, the specific area division of the image can be effectively completed.

[0038] Identify the center point of the partitioned area and record it as the reference point. Then, identify the feature point in the partitioned area that is farthest from the reference point (i.e., the straight-line distance). Identify the feature line between the reference point and the feature point. Record the pixels on the feature line as the line pixel points. Arrange the line pixel points in order from near to far from the reference point to determine a set of line pixel point arrangements.

[0039] Then, the edge pixels in the partitioned area are processed in the same way to determine the different line pixel permutations associated with different edge pixels (the permutation set associated with the feature point is the longest, and the permutation set corresponding to the edge point at a vertical distance is the shortest).

[0040] Pixels at the same sorted position are selected from a set of connected pixel points, and pixel values ​​associated with the corresponding pixel points at the same sorted position are confirmed, and feature verification is performed on the confirmed pixel values: a group of pixel values ​​is randomly selected as the intermediate pixel value Zh, and the clustering interval associated with the pixel value is confirmed using: [Zh-Y1, Zh+Y1], where Y1 is a preset value, generally 10-20, which is prepared in advance by the operator based on experience, and the clustering pixel value belonging to this clustering interval is used as the clustering pixel value of this pixel value, and different pixel values ​​are selected as intermediate pixel values ​​in turn, and the clustering pixel values ​​are gradually determined, and a group of clustering intervals with the largest total number of clustering pixel values ​​is selected as the standard interval (different selection processes have different numbers of clustering pixel values), based on the confirmed standard interval, pixel values ​​that do not belong to this standard interval are calibrated as abnormal pixel values, and pixel points associated with the abnormal pixel values ​​are calibrated as abnormal pixel points;

[0041] And several pixels at different sorting positions in several connected pixel point arrangement sets are calibrated as abnormal pixels.

[0042] Specific, combined Figure 2 Based on the confirmed partition area, the center point inside it is confirmed as the corresponding reference point. After the reference point is determined, the farthest feature point and the nearest feature point associated with it can be confirmed, so that the corresponding line B and line A can be confirmed. The number of associated pixel points on the line B is the largest, and the number of associated pixel points on the line A is the smallest.

[0043] Then, based on the confirmed line A and line B, the relevant pixel points at the same sorting position can be confirmed. When the relevant pixel points at the corresponding sorting position do not exist in the corresponding pixel point arrangement set, they will be ignored. The pixel points associated with other arrangement sets can be numerically confirmed. In other words, there are only four groups of corner points in this partitioned area. Then, for the pixel point at the last position in the line pixel point arrangement set associated with the farthest feature point, there are only four pixel points at the corresponding sorting position. That is, the pixel values ​​of the four pixel points are verified in the final stage.

[0044] Among them, the image different area calibration end calibrates the adjacent abnormal pixel points for the several abnormal pixel points calibrated in the image of the relevant area, confirms the specific area associated with the adjacent abnormal pixel points, records the associated specific area as the abnormal area, and calibrates it in the image of the relevant area;

[0045] Among them, the feature verification processing end determines the spatial position features associated with different pixel points in the abnormal area based on several abnormal areas marked in the relevant area image, locks the feature value of the abnormal area based on the confirmed spatial position features, and transmits the locked feature value to the area calibration display end. The specific method of determining the spatial position features is as follows:

[0046] Confirm the spatial position of the camera, confirm the vertical point of this spatial position from the confirmed relevant area image, and record the pixel point associated with this vertical point as the feature check point. The feature check point is the vertical point, and the spatial position of the corresponding camera is the corresponding spatial positioning point. Combined with the specific image captured, it is possible to confirm the vertical point associated with this spatial positioning point vertically downward, and record the pixel value associated with the feature check point as Xd, and the vertical distance associated with the feature check point as Jl;

[0047] Then the pixel values ​​associated with different pixel points in the abnormal area are calibrated as X k , where k represents different pixels. When the object is closer to the camera, it occupies a relatively large number of pixels in the image, and the actual object area corresponding to each pixel is smaller. Therefore, the object's details can be captured more finely, and the object features reflected by the pixel value are more accurate and richer. Therefore, it can be said that the closer the distance, the higher the pixel value associated with the pixel. Use: |Xd-X k |×C1+Jl=J k Confirm the feature distance J associated with the corresponding pixel point k , where C1 is a preset fixed coefficient factor, the specific value of which is determined by the operator based on experience, and C1 is generally between 5 and 10;

[0048] Confirm the horizontal distance between the corresponding pixel point and the feature verification point from the image of the relevant area (the distance cannot be confirmed within the image, and the image is a two-dimensional plane, so only a horizontal distance can be confirmed). Based on the determined horizontal distance and the location of the corresponding pixel point, calibrate the pixel point in the same plane as the feature verification point. The calibrated position is the virtual position. Then, a set of specific vertical lines perpendicular to the virtual position are constructed and calibrated as virtual vertical lines. Confirm the straight-line distance between different points and the spatial position of the shooting camera from the virtual vertical line. Confirm the straight-line distance = J k The point is used as the spatial position feature point of this pixel;

[0049] Then confirm the spatial position feature points associated with different pixel points in the abnormal area in turn;

[0050] Based on the spatial position feature points associated with different pixel points in the same abnormal area, a set of lowest points and a set of highest points are confirmed (by constructing a set of horizontal base planes, the height characteristics of the corresponding spatial position feature points can be confirmed, thereby determining the lowest point and the highest point). Then, the horizontal base plane where the lowest point is located is confirmed, and the vertical distance associated with the highest point and the horizontal base plane is locked. The locked vertical distance is used as the characteristic value of this abnormal area.

[0051] Specifically, the so-called characteristic value is the height difference of the corresponding abnormal area. The value is confirmed by the correlation characteristics between the corresponding highest point and the lowest point in the corresponding abnormal area, so that the vertical distance associated with different spatial points can be confirmed, and thus the height difference associated with the corresponding abnormal area can be confirmed.

[0052] Among them, the regional calibration display terminal confirms the disease of different abnormal areas based on the different characteristic values ​​associated with different abnormal areas, locks the diseased areas and displays them. The specific method of disease confirmation is:

[0053] The different feature values ​​associated with different abnormal areas in the relevant area image are calibrated as TZ q , where q represents different abnormal regions, will satisfy: TZ q Abnormal areas with a value ≥ Y2 are calibrated as diseased areas. Otherwise, no calibration is performed. Y2 is a preset value, and its specific value is determined by the operator based on experience. It is generally between 30-40 cm, and its specific value is determined in combination with the specific road conditions.

[0054] Some of the data in the above formulas are dimensionless and numerically calculated. Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0055] The above embodiments are only used to illustrate the technical method of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A road hazard intelligent identification and early warning system, characterized in that: include: The image feature processing end grayscales the corresponding image based on the acquired image of the relevant area, confirms the grayscale image, and then calibrates the abnormal pixels based on the pixel value features associated with different points in the grayscale image; The image different area calibration end calibrates the adjacent abnormal pixel points for the several abnormal pixel points calibrated in the image of the relevant area, and confirms that the specific area associated with the adjacent abnormal pixel points is recorded as the abnormal area; The feature verification processing end determines the spatial position features associated with different pixels in the abnormal area based on several abnormal areas marked in the relevant area image. Based on the confirmed spatial position features, the feature value of the abnormal area is locked. The specific method is as follows: Confirm the spatial position of the camera, confirm the vertical point of this spatial position from the confirmed relevant area image, record the pixel point associated with this vertical point as the feature check point, record the pixel value associated with the feature check point as Xd, and record the vertical distance associated with the feature check point as Jl; Then the pixel values ​​associated with different pixel points in the abnormal area are calibrated as X k , where k represents different pixel points, using: |Xd-X k |×C1+Jl=J k Confirm the feature distance J associated with the corresponding pixel point k , where C1 is the preset fixed coefficient factor; Determine the horizontal distance between the corresponding pixel point and the feature verification point in the image of the relevant area. Based on the determined horizontal distance and the location of the corresponding pixel point, calibrate the pixel point in the same plane as the feature verification point. The calibrated position is the virtual position. Then, construct a set of specific vertical lines perpendicular to the virtual position and calibrate them as virtual vertical lines. Determine the straight-line distance between different points and the spatial position of the camera from the virtual vertical lines. Confirm that the straight-line distance = J k The point is used as the spatial position feature point of this pixel; Then confirm the spatial position feature points associated with different pixel points in the abnormal area in turn; Based on the spatial position feature points associated with different pixel points in the same group of abnormal areas, a group of lowest points and a group of highest points are confirmed, and then the horizontal base plane where the lowest points are located is confirmed, and the vertical distance associated with the highest point and the horizontal base plane is locked. The locked vertical distance is used as the characteristic value of this abnormal area.

2. The road hazard intelligent identification and early warning system according to claim 1, characterized in that: Also includes: The regional image acquisition end confirms the relevant regional images acquired by the image survey vehicle and transmits the confirmed relevant regional images to the image feature processing end.

3. The road hazard intelligent identification and early warning system according to claim 1, characterized in that: The specific method of calibrating abnormal pixels at the image feature processing end is as follows: Gray-scale processing is performed on the acquired image of the relevant area, the gray-scale image is confirmed, and then the confirmed gray-scale image is divided into equal areas to confirm a number of different partitioned areas; Determine the center point of the partitioned area and record it as the reference point. Then, determine the feature point farthest from the reference point in the partitioned area. Determine the feature line between the reference point and the feature point. Record the pixels on the feature line as the line pixel points. Arrange the line pixel points in order from near to far from the reference point to determine a set of line pixel point arrangement. Then, the edge pixels in the partitioned area are processed in the same manner to determine the different line pixel arrangement sets associated with different edge pixels; Pixel points at the same sorting position are selected from a set of connected pixel points, pixel values ​​associated with the corresponding pixel points at the same sorting position are confirmed, and feature verification is performed on the confirmed pixel values: a group of pixel values ​​is randomly selected as the intermediate pixel value Zh, and the clustering interval associated with the pixel value is confirmed using: [Zh-Y1, Zh+Y1], where Y1 is a preset value, and the pixel value belonging to this clustering interval is used as the clustering pixel value of this pixel value, different pixel values ​​are selected in turn as the intermediate pixel values, and the clustering pixel values ​​are gradually determined, and a group of clustering intervals with the largest total number of clustering pixel values ​​is selected as the standard interval, and based on the confirmed standard interval, the pixel values ​​that do not belong to this standard interval are calibrated as abnormal pixel values, and the pixel points associated with the abnormal pixel values ​​are calibrated as abnormal pixel points; And several pixels at different sorting positions in several connected pixel point arrangement sets are calibrated as abnormal pixels.

4. The road hazard intelligent identification and early warning system according to claim 3 is characterized in that: When the area is equally divided, a preset equal division template exists, and the equal division template contains a number of equal division grids of the same area, and each equal division grid corresponds to a group of partitioned areas.

5. The road hazard intelligent identification and early warning system according to claim 1 is characterized in that: Also includes: The area calibration display terminal confirms the disease of different abnormal areas based on the different characteristic values ​​associated with different abnormal areas, locks the diseased areas and displays them.

6. The road hazard intelligent identification and early warning system according to claim 5, characterized in that: The specific method of locking the diseased area at the area calibration display terminal is as follows: The different feature values ​​associated with different abnormal areas in the relevant area image are calibrated as TZ q , where q represents different abnormal regions, will satisfy: TZ q Abnormal areas with a value of ≥Y2 are marked as diseased areas.

7. The road hazard intelligent identification and early warning system according to claim 6, characterized in that: Will not satisfy: TZ q For abnormal areas ≥Y2, no calibration is performed and Y2 is the preset value.

Images