CN-122024059-A - Highway pavement quality detection method and system

CN122024059ACN 122024059 ACN122024059 ACN 122024059ACN-122024059-A

Abstract

The invention relates to the technical field of image processing, in particular to a highway pavement quality detection method and system, comprising the following steps of constructing a minimum interference frame group based on an image time sequence in a travelling direction, wherein the minimum interference frame group is selected from the image time sequence by a minimum inter-frame brightness gradient change principle; performing regional heterogeneous reconstruction on the minimum interference frame group to generate a candidate distortion region reflecting local texture stability and edge distortion characteristics, introducing multi-scale structural entropy analysis to quantize the inflection point characteristics into structural instability indexes, generating a pavement distortion response map by fusing the candidate distortion region and the corresponding structural instability indexes, and identifying pavement defect regions according to gradient direction consistency, form closure degree and the structural instability indexes of each region in the pavement distortion response map. Compared with the traditional method relying on manual interpretation or single index statistics, the method provided by the invention has the advantage that the accuracy and consistency are obviously improved.

Inventors

XU DONGMEI
LI HONGLE
LI WANYOU

Assignees

山东路泰公路工程有限公司

Dates

Publication Date: 20260512
Application Date: 20260130

Claims (10)

1. The highway pavement quality detection method is characterized by comprising the following steps of: S1, acquiring image frames of a road surface to be detected, and constructing a minimum interference frame group based on an image time sequence in the advancing direction, wherein the minimum interference frame group is selected from the image time sequence by a minimum principle of inter-frame brightness gradient change; S2, carrying out regional heterogeneous reconstruction on the minimum interference frame group, wherein the regional heterogeneous reconstruction comprises carrying out position pixel clustering based on pixel position information of each image frame in the same frame group, generating a candidate distortion region reflecting local texture stability and edge distortion characteristics, introducing multi-scale structural entropy analysis, calculating structural chaos entropy values of texture trend from pixel level to block level under multiple scales in the candidate distortion region, extracting inflection point characteristics of the structural chaos entropy values changing along with the scales, and quantifying the inflection point characteristics into structural instability indexes, and the pavement distortion response map is generated by fusing the candidate distortion region and the corresponding structural instability indexes; and S3, identifying a pavement defect area according to the gradient direction consistency, the form closure degree and the structural instability index of each area in the pavement distortion response chart, and comprehensively marking and outputting the pavement quality grade.
2. The method for detecting the quality of the road surface according to claim 1, wherein in the step S1, the image frames of the road surface to be detected are continuously acquired at fixed time or displacement intervals along the advancing direction by an image acquisition unit carried on a detection vehicle to form the image time sequence, each frame image in the image time sequence is subjected to gray-scale processing, a brightness gradient amplitude image of the image is calculated, and brightness gradient amplitude variation between adjacent frames in the image time sequence is sequentially calculated, wherein the variation is measured by a difference norm of gradient amplitudes of corresponding pixels.
3. The method according to claim 2, wherein S1 further comprises setting a variation threshold, selecting a continuous frame sub-sequence in the image time sequence, such that the brightness gradient variation between all adjacent frames in the frame sub-sequence is lower than the variation threshold, and the frame number included in the frame sub-sequence reaches a preset number, and constructing the frame sub-sequence as the minimum interference frame group.
4. The method for detecting the quality of a highway pavement according to claim 1, wherein the generation of the candidate distortion region specifically comprises: performing pixel coordinate alignment and normalization on each image frame in the minimum interference frame group; Based on the aligned image frames, aiming at each position in a pavement preset grid, aggregating pixels from all frames at corresponding positions to form a set of pixels at the positions, performing unsupervised clustering based on gray values, gradient directions and neighborhood contrast of the pixels at the positions, and marking a set of which the integral attribute difference between a clustering center and the set exceeds a preset attribute difference threshold value as a heterogeneous pixel set.
5. The method according to claim 4, wherein S2 further comprises merging and morphological processing the heterogeneous pixel sets that are spatially adjacent and have similar properties in a road space domain to generate spatially continuous candidate distortion regions, performing multi-scale structural entropy analysis for each candidate distortion region, defining a plurality of analysis scales from a single pixel, a local block to an entire region, counting a distribution of gradient directions of all pixels in the region at each analysis scale, and calculating shannon entropy values of the distribution as structural confusion entropy values at the corresponding analysis scale.
6. The method for detecting the quality of the highway pavement according to claim 5, wherein the quantification of the structural instability index comprises the steps of drawing a change curve of the entropy value of the structural confusion degree of the candidate distortion region along with the increase of an analysis scale, identifying an inflection point of which the change rate of the entropy value changes from rapid to slow, and carrying out linear combination on the scale value corresponding to the inflection point, the average slope ratio of the curve before and after the inflection point and the entropy value at the inflection point, so as to calculate the structural instability index of the corresponding region; And taking each candidate distortion region as a basic image layer, taking the corresponding structural instability index as an intensity image layer overlapped on the region, and carrying out weighted fusion to generate the pavement distortion response image.
7. The method according to claim 1, wherein the step S3 comprises extracting boundary contours of each candidate distortion region from the road distortion response map, and calculating gradient direction consistency of edge pixels in each boundary contour, and simultaneously calculating a ratio of an area of each contour to an area of a minimum circumscribing convex polygon as a morphological closing degree index.
8. The method for detecting the quality of the road surface according to claim 7, wherein the step S3 further comprises the steps of establishing a defect recognition model, wherein the input characteristics of the defect recognition model are gradient direction consistency, form closure degree index and structure instability index of the corresponding region, and outputting the defective region of the road surface by using a pre-trained support vector machine as the defect recognition model.
9. The method for detecting the quality of the road surface according to claim 8, wherein all the identified pavement defect areas are sorted and classified into a plurality of defect levels according to the numerical values of the structural instability indexes of the pavement defect areas, the total defect areas of the defect levels and the spatial distribution density of the defect areas in the detected road sections are counted, and the pavement quality level marks are output to be good, qualified, inferior and dangerous four levels based on the highest level, the total defect area occupation ratio and the spatial distribution density of all the defect areas in the detected road sections through calculation according to a predefined comprehensive grading rule.
10. A road surface quality detection system for implementing a road surface quality detection method according to any one of claims 1-9, characterized by comprising the following modules: The image acquisition module is used for acquiring an image time sequence of the road surface to be detected; The image preprocessing module is used for constructing a least interference frame group from the image time sequence based on the principle of minimum inter-frame brightness gradient change; The region reconstruction and structure entropy analysis module is used for performing co-located pixel clustering on the minimum interference frame group to generate a candidate distortion region, and performing multi-scale structure entropy analysis to extract a structure instability index and construct a pavement distortion response diagram; The defect identification and quality grade evaluation module is used for identifying a defect area according to the gradient direction consistency, the form closure degree and the structural instability index of each candidate area in the distortion response diagram and outputting a pavement quality grade result.

Description

Highway pavement quality detection method and system Technical Field The invention relates to the technical field of image processing, in particular to a highway pavement quality detection method and system. Background In the maintenance process of road traffic infrastructure, detection and evaluation of road surface quality are a vital task, and are directly related to driving safety, road life and scientificity of maintenance decision. At present, the road surface quality detection mainly depends on manual inspection and visual judgment, the method relies on manually carrying detection tools to carry out visual inspection identification or manual recording on apparent defects (such as cracks, pits, sinkers and the like) of the road surface, and the method has the advantages of low cost, extremely low detection efficiency, strong subjectivity of data, difficulty in realizing high-frequency and large-scale real-time detection and inadaptability to the development requirements of informatization and intellectualization of the current road maintenance. In recent years, a part of high-grade highways are introduced with laser scanning equipment or three-dimensional structure optical modules for acquiring road surface elevation data and constructing a three-dimensional structure model. However, such equipment is expensive, has high requirements on the working environment, and has insufficient detection capability for early recessive diseases such as surface color difference, texture disturbance and the like. In addition, the data processing flow is complex, high-performance computing resources are relied on, and the method is difficult to popularize and apply in basic road sections and large-scale highway networks. Compared with the traditional method, the road surface detection technology based on image processing is paid attention to widely due to the advantages of low cost, flexible deployment, strong adaptability and the like. Such methods typically employ an industrial camera, vehicle-mounted camera, or mobile terminal to acquire a sequence of road images, identifying potential defect areas by means of image enhancement, edge detection, texture analysis, and the like. However, the existing image detection technology still has the defects that in an actual road environment, due to factors such as illumination mutation, camera shake, motion blur and the like, the stability of an image sequence is poor, and false detection or omission is extremely easy to generate. Most methods rely on single frame images, lack a data fusion mechanism of multi-view and multi-time nodes, and are difficult to identify hidden diseases such as light distortion, boundary cracks and the like. Disclosure of Invention The invention provides a road surface quality detection method and a road surface quality detection system, which are integrated with multi-frame image information, have structural stability modeling capability and support multi-level quality grading, so as to break through the bottleneck of the prior art in the aspects of detection precision, stability and generalizability and realize more efficient and intelligent road quality detection and maintenance. A highway pavement quality detection method comprises the following steps: S1, acquiring image frames of a road surface to be detected, and constructing a minimum interference frame group based on an image time sequence in the advancing direction, wherein the minimum interference frame group is selected from the image time sequence by a minimum principle of inter-frame brightness gradient change so as to weaken dynamic interference of vehicle shadows and instantaneous light spots; S2, carrying out regional heterogeneous reconstruction on the minimum interference frame group, wherein the regional heterogeneous reconstruction comprises carrying out position pixel clustering based on pixel position information of each image frame in the same frame group, generating a candidate distortion region reflecting local texture stability and edge distortion characteristics, introducing multi-scale structural entropy analysis, calculating structural chaos entropy values of texture trend from pixel level to block level under multiple scales in the candidate distortion region, extracting inflection point characteristics of the structural chaos entropy values changing along with the scales, and quantifying the inflection point characteristics into structural instability indexes, and the pavement distortion response map is generated by fusing the candidate distortion region and the corresponding structural instability indexes; and S3, identifying a pavement defect area according to the gradient direction consistency, the form closure degree and the structural instability index of each area in the pavement distortion response chart, and comprehensively marking and outputting the pavement quality grade. Optionally, in the step S1, the image frames of the road pavement to be detected are c