CN115272323A - Data intelligent regulation and control acquisition method for traffic engineering pavement quality detection - Google Patents
Data intelligent regulation and control acquisition method for traffic engineering pavement quality detection Download PDFInfo
- Publication number
- CN115272323A CN115272323A CN202211185973.8A CN202211185973A CN115272323A CN 115272323 A CN115272323 A CN 115272323A CN 202211185973 A CN202211185973 A CN 202211185973A CN 115272323 A CN115272323 A CN 115272323A
- Authority
- CN
- China
- Prior art keywords
- image
- segmentation
- road surface
- curve
- abnormal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 41
- 238000001514 detection method Methods 0.000 title claims abstract description 13
- 230000011218 segmentation Effects 0.000 claims abstract description 71
- 230000002159 abnormal effect Effects 0.000 claims abstract description 69
- 238000001228 spectrum Methods 0.000 claims abstract description 55
- 230000009467 reduction Effects 0.000 claims abstract description 37
- 238000013528 artificial neural network Methods 0.000 claims abstract description 20
- 230000003595 spectral effect Effects 0.000 claims description 13
- 230000005856 abnormality Effects 0.000 claims description 11
- 230000001276 controlling effect Effects 0.000 claims description 10
- 230000001105 regulatory effect Effects 0.000 claims description 9
- 239000000126 substance Substances 0.000 claims description 6
- 238000005192 partition Methods 0.000 claims description 3
- 238000010183 spectrum analysis Methods 0.000 abstract 1
- 238000010586 diagram Methods 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 4
- 238000005070 sampling Methods 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 150000001875 compounds Chemical class 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000002372 labelling Methods 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 230000000737 periodic effect Effects 0.000 description 2
- 238000011176 pooling Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 241000239290 Araneae Species 0.000 description 1
- 206010040844 Skin exfoliation Diseases 0.000 description 1
- 239000010426 asphalt Substances 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000035618 desquamation Effects 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30248—Vehicle exterior or interior
- G06T2207/30252—Vehicle exterior; Vicinity of vehicle
- G06T2207/30256—Lane; Road marking
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Quality & Reliability (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention relates to the field of intelligent regulation, in particular to an intelligent regulation and acquisition method for data used for detecting the road surface quality of traffic engineering. Acquiring a frequency spectrum image of a pavement gray level image; constructing a dimensionality reduction sequence of the frequency spectrum image, fitting a curve, and calculating a curve slope between each element and the last element in the curve; calculating the size of a segmentation frame, and performing sliding segmentation on the frequency spectrum image to obtain a plurality of segmentation areas; calculating the linear abnormal degree of each segmentation region; setting the element value of the segmentation area with the linear abnormal degree larger than the threshold value as 1 to obtain a road surface abnormal binary image, and controlling a vehicle-mounted rear camera to acquire an abnormal area image; and inputting the area image acquired by the vehicle-mounted rear camera into the neural network, and judging the pavement quality by utilizing the pavement abnormal category output by the neural network. According to the invention, the rear camera is controlled to collect abnormal areas through the spectrum analysis result of the image collected by the front camera, so that the high-precision detection of the pavement quality can be realized.
Description
Technical Field
The invention relates to the field of intelligent regulation, in particular to an intelligent regulation and acquisition method for data used for detecting the road surface quality of traffic engineering.
Background
The main forms of pavement diseases affecting quality include cracks, pot holes, desquamation, subsidence and rutting, and in the prior art, the identification and detection are usually carried out by using a neural network mode, but the mode needs a large amount of data with manual labels for supervision and training. When the images are collected, the collected images are mostly normal images, the sizes of cracks in the images are sometimes small and difficult to identify, the abnormality of the road surface is identified in advance by using an image processing mode, if all the images are collected and each image is comprehensively and carefully analyzed, the calculation cost of a calculation unit is relatively high.
Aiming at the problems, the invention provides an intelligent data regulation and collection method for detecting the road surface quality of traffic engineering, which is used for quickly judging abnormality from a frequency domain, then regulating a rear camera and collecting images which are more likely to be abnormal.
Disclosure of Invention
The invention provides an intelligent data regulation and acquisition method for traffic engineering pavement quality detection, which aims to solve the existing problems. The method comprises the following steps: acquiring a frequency spectrum image of a pavement gray level image; constructing a dimensionality reduction sequence of the frequency spectrum image, fitting a curve, and calculating a curve slope between each element and the last element in the curve; calculating the size of a segmentation frame, and performing sliding segmentation on the frequency spectrum image to obtain a plurality of segmentation areas; calculating the linear abnormal degree of each segmentation region; setting the element value of the segmentation area with the linear abnormal degree larger than the threshold value as 1 to obtain a road surface abnormal binary image, and controlling a vehicle-mounted rear camera to acquire an abnormal area image; and inputting the area image acquired by the vehicle-mounted rear camera into a neural network, and judging the road surface quality by using the road surface abnormal category output by the neural network.
According to the technical means provided by the invention, fourier transformation is carried out on the image, the abnormity is rapidly judged from the frequency domain, and then the rear camera is controlled to collect the abnormal area according to the estimated abnormal area, so that the power consumption of the system is reduced on the whole, the image does not need to be collected all the time, the high-precision abnormal area image is obtained to be identified, the identification precision of a neural network is improved, and the high-precision detection of the pavement quality is realized.
The invention adopts the following technical scheme that a data intelligent regulation and control acquisition method for traffic engineering pavement quality detection comprises the following steps:
and collecting the pavement gray level image to obtain a frequency spectrum image of the pavement gray level image.
Constructing a dimensionality reduction sequence of the spectral image according to the frequency energy mean value of each row of elements in the spectral image; and performing curve fitting on the dimensionality reduction sequence of the frequency spectrum image, and calculating the slope of the curve between each element and the last element in the curve.
And calculating the size of a segmentation frame according to the sequence number of the element corresponding to the maximum slope value of the curve between the last elements in the dimension reduction sequence, and performing sliding segmentation on the frequency spectrum image by using the obtained size of the segmentation frame to obtain a plurality of segmentation areas.
And calculating the linear abnormal degree of each segmentation region according to the curve slope between each element and the last element in the dimensionality reduction sequence curve corresponding to each segmentation region.
And setting the element values of all the segmentation areas with the linear abnormal degrees larger than the threshold value as 1 to obtain the road surface abnormal binary image.
Dividing the road surface abnormal binary image into a plurality of acquisition areas, and controlling the vehicle-mounted rear camera of the corresponding area to acquire the image when the element value in the corresponding acquisition area is 1.
And inputting the image acquired by the vehicle-mounted rear camera into a neural network, and judging the road surface quality by using the road surface abnormal category output by the neural network.
Furthermore, a method for intelligently regulating and acquiring data for detecting the road surface quality of traffic engineering, which comprises the following steps of constructing a dimensionality reduction sequence of a frequency spectrum image according to the frequency energy mean value of each row of elements in the frequency spectrum image:
converting the road surface gray level image with the size of C multiplied by K into a frequency spectrum image by utilizing Fourier transform;
and taking the center of the frequency spectrum as a starting point, acquiring the frequency energy mean value of each row of elements in the right half part of the frequency spectrum image, and constructing a dimension reduction sequence of the frequency spectrum image according to the frequency energy mean value of each row of elements.
Further, the method for intelligently regulating and acquiring data for detecting the road surface quality of the traffic engineering comprises the following steps of:
calculating the slope of the curve between each element and the last element in the curve, and acquiring the average slope of the slopes of all the elements;
obtaining the difference degree of each element according to the square of the difference value between the slope of the curve and the average slope between each element and the last element, and obtaining the serial number value of the element corresponding to the maximum difference degree in the dimension reduction sequence;
and calculating the size of the segmentation frame according to the sequence number value of the element in the dimension reduction sequence and the length of the frequency spectrum image.
Further, an intelligent data regulation and acquisition method for detecting the road surface quality of traffic engineering is characterized in that a frequency spectrum image is subjected to sliding segmentation, and a plurality of segmented regions are obtained by the method comprising the following steps:
obtaining a segmentation direction according to a curve slope and an average slope between an element corresponding to the maximum difference degree and the last element;
wherein the content of the first and second substances,
representing a dividing direction parameter, concave transverse
Convex in the longitudinal direction
，
Representing the slope of the curve between the element corresponding to the greatest degree of difference and the last element,
representing the average slope, and s represents the serial number of the element corresponding to the maximum difference degree in the dimension reduction sequence;
and performing sliding segmentation on the frequency spectrum image in the corresponding segmentation direction by using the segmentation frame to obtain a plurality of segmentation areas.
Further, a data intelligent regulation and collection method for traffic engineering pavement quality detection, the method for calculating the linear abnormal degree of each segmentation area comprises the following steps:
the expression for calculating the linear abnormality degree is as follows:
wherein the content of the first and second substances,
which is indicative of the degree of the linear anomaly,
representing the slope of the curve between the ith element and the last element in the dimensionality reduction sequence,
which represents the average slope of the light beam,
is the number of columns of elements in the spectral image;
and acquiring all corresponding elements of each segmentation region in the dimension reduction sequence, acquiring the slope of a curve between each element in each segmentation region and the last element of the region, and calculating the linear abnormal degree of each segmentation region by using an expression for calculating the linear abnormal degree.
Further, an intelligent data regulation and collection method for detecting the road surface quality of traffic engineering is characterized in that a method for dividing a road surface abnormal binary image into a plurality of collection areas comprises the following steps:
acquiring acquisition time according to the number of lines of elements in the road surface abnormal binary image and the running speed of a vehicle where the vehicle-mounted camera is located, and longitudinally dividing the road surface abnormal binary image into a plurality of areas according to the acquisition frequency and the acquisition time of the rear camera;
and horizontally dividing the road surface abnormal binary image into a plurality of acquisition areas according to the number of the rear vehicle-mounted cameras.
Further, the method for intelligently regulating and acquiring the data for detecting the road surface quality of the traffic engineering is characterized in that the method for controlling the vehicle-mounted rear camera of the corresponding area to acquire the image of the area comprises the following steps:
acquiring all elements with element values of 1 in a road surface abnormal binary image, and controlling a rear camera to acquire images when the elements with the element values of 1 appear in an acquisition area corresponding to the vehicle-mounted rear camera;
and when the element with the element value of 1 does not appear in the acquisition area corresponding to the vehicle-mounted rear camera, the rear camera does not acquire the image.
The invention has the beneficial effects that: according to the technical means provided by the invention, the images are subjected to Fourier transform, the abnormity is quickly judged from the frequency domain, and the rear camera is controlled to collect the abnormal area according to the estimated abnormal area, so that the power consumption of the system is reduced on the whole, the images do not need to be collected anytime and anywhere, the high-precision abnormal area images are obtained to be identified, the identification precision of a neural network is improved, and the high-precision detection of the pavement quality is realized.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the embodiments or the description of the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a schematic structural diagram of a data intelligent regulation and acquisition method for traffic engineering pavement quality detection according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a normal road surface curve fitting according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a curve fitting of a lateral pavement crack according to an embodiment of the invention;
FIG. 4 is a schematic diagram of a curve fitting of longitudinal cracks of a pavement according to an embodiment of the present invention;
fig. 5 is a schematic diagram of a road surface cobweb-shaped crack curve fitting according to an embodiment of the invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention.
As shown in fig. 1, a schematic structural diagram of a data intelligent regulation and acquisition method for detecting road surface quality of traffic engineering according to an embodiment of the present invention is provided, which includes:
101. and collecting the pavement gray level image to obtain a frequency spectrum image of the pavement gray level image.
The invention utilizes a corresponding acquisition system to acquire road image information, and the acquisition system comprises: the system comprises a vehicle, a front camera, a rear camera and a lighting device, and then collected image data are transmitted to a corresponding cloud server through a data transmission device, and then the image is processed by running a corresponding calculation process in the server, so that a road pavement quality detection result is obtained.
The image collected by the front camera is recorded as
The acquired image of the rear camera is recorded as
The number of the front cameras is higher than that of the front cameras, and the number relation is as follows:
。
i.e. the number of rear cameras being front cameras
The image of the double, i.e. front, camera can be divided into
The angle of view of the rear camera.
The front camera is a high-frequency continuous acquisition range with a wide angle of view, and the rear camera is a self-adaptive adjusting camera.
After the front-facing camera collects the images, the images are processed by utilizing Fourier transform, and corresponding frequency spectrum images can be obtained. The acquisition process of the frequency spectrum comprises the following steps:
and carrying out graying operation on the collected RGB image to obtain a grayscale image.
And performing two-dimensional discrete Fourier transform on the gray-scale image to obtain a magnitude spectrum, and then adjusting four quadrants to obtain a centralized magnitude spectrum, namely a spectrogram.
Abnormal information, types and positions of cracks can be analyzed rapidly in the frequency spectrum image, then a camera is adjusted to conduct sampling, and then high-accuracy identification is conducted.
In a normal road surface image, asphalt particles are uniformly distributed, and the response of low frequency and high frequency on a frequency spectrum is smooth and gradually changed. And after the abnormality occurs on the road surface, the response of the low-frequency area is more severe.
102. Constructing a dimensionality reduction sequence of the spectral image according to the frequency energy mean value of each row of elements in the spectral image; and performing curve fitting on the dimensionality reduction sequence of the frequency spectrum image, and calculating the slope of the curve between each element and the last element in the curve.
The spectrum is symmetrical about the center, and the size of the spectrum is consistent with the size of the corresponding processed image, denoted as [ C, K ].
In the spectrum, a region close to the center is a low-frequency region, a surrounding region far away from the center is a high-frequency region, and each element in the spectrum represents the participation degree of a periodic texture corresponding to the frequency and the direction in a current image (the image is formed by overlapping a plurality of periodic textures), so that in order to represent the information of the whole two-dimensional spectrum, the dimension reduction processing needs to be carried out on the information to obtain the one-dimensional representation of the spectrum:
taking the center point of the frequency spectrum as a starting point, taking C/2 as a sequence length, obtaining the mean value of each column of the right half part of the image, wherein the calculation formula of the corresponding frequency energy mean value is as follows:
in the formula (I), the compound is shown in the specification,
denotes the first
The number of columns and the mean value,
the element values of j rows and i columns in the frequency spectrum are shown, and K represents the total K rows of elements.
The sum of the values of all elements per column and/or the number of elements per column = mean value of the current frequency
Thus, a one-dimensional sequence corresponding to the spectrum can be obtained:
for the one-dimensional sequence, the lower the frequency in the sequence is, the higher the frequency in the sequence is, and therefore the one-dimensional sequence is a dimension-reduced sequence of the spectrum image.
The method for constructing the dimensionality reduction sequence of the spectral image according to the frequency energy mean value of each row of elements in the spectral image comprises the following steps:
converting the road surface gray level image with the size of C multiplied by K into a frequency spectrum image by utilizing Fourier transform;
and taking the center of the frequency spectrum as a starting point, acquiring the frequency energy mean value of each row of elements in the right half part of the frequency spectrum image, and constructing a dimension reduction sequence of the frequency spectrum image according to the frequency energy mean value of each row of elements.
The curve fitting of the dimension reduction sequence of the frequency spectrum images of different road surfaces is shown in fig. 2, fig. 3, fig. 4 and fig. 5, the given curve images are scattered point data and are not fitted, so that the degree of sawtooth fluctuation is relatively large, and according to the images, different road surface abnormalities can cause changes of different degrees of high frequency and low frequency, so that the relative relation of the whole dimension reduction sequence is changed.
103. And calculating the size of a segmentation frame according to the sequence number of the element corresponding to the maximum slope value of the curve between the last elements in the dimension reduction sequence, and performing sliding segmentation on the frequency spectrum image by using the obtained size of the segmentation frame to obtain a plurality of segmentation areas.
The method for calculating the size of the segmentation frame comprises the following steps:
calculating the slope of the curve between each element and the last element in the curve, and acquiring the average slope of the slopes of all the elements;
firstly, the slope determined by each point and the last point is obtained
：
Wherein, the first and the second end of the pipe are connected with each other,
for the ith data in the sequence,
for the C/2-i data in the sequence,
is the average slope.
The more inconsistent the slope, the greater the degree of abnormality in the image. The dimension reduction sequence of the pavement without cracks is more in line with the linear relation.
Obtaining the difference degree of each element according to the square of the difference value between the slope of the curve and the average slope between each element and the last element, and obtaining the serial number value of the element corresponding to the maximum difference degree in the dimension reduction sequence;
obtaining the sequence number corresponding to the maximum difference in the sequence:
the greatest difference is obtained
Corresponding serial number
I.e. by
Is/are as follows
A value of i.e.
(the position where the difference from the linear fit is the greatest).
And calculating the size of the segmentation frame according to the sequence number value of the element in the dimension reduction sequence and the length of the frequency spectrum image.
Calculating the size of the divided frame
：
In the formula (I), the compound is shown in the specification,
c is the long dimension of the image for the scale factor.
In the place where the slope difference is the largest, the degree of the high frequency and low frequency is such that the lower the frequency, the larger the division ratio is, and the higher the frequency, the smaller the division ratio is.
The method for performing sliding segmentation on the frequency spectrum image to obtain a plurality of segmentation areas comprises the following steps:
obtaining a segmentation direction according to a curve slope and an average slope between an element corresponding to the maximum difference degree and the last element;
wherein, the first and the second end of the pipe are connected with each other,
representing a dividing direction parameter, concave transverse
Convex in the longitudinal direction
，
Representing the slope of the curve between the element corresponding to the greatest degree of difference and the last element,
representing the average slope, and s represents the serial number of the element corresponding to the maximum difference degree in the dimension reduction sequence;
and performing sliding segmentation on the frequency spectrum image in the corresponding segmentation direction by using the segmentation frame to obtain a plurality of segmentation areas.
If the partition is horizontal, the size of the partition frame is
；
If the segmentation is performed in the longitudinal direction, the size of the division frame is
；
Next, the divided frames are overlapped and divided, and the divided frames are moved at a certain pitch, the moving distance is a corresponding proportion of the size of the frame corresponding to the moving direction, and the empirical value is 0.5 (i.e., half frame is moved each time).
104. And calculating the linear abnormal degree of each segmentation region according to the curve slope between each element and the last element in the dimensionality reduction sequence curve corresponding to each segmentation region.
The method for calculating the linear abnormal degree of each segmentation region comprises the following steps:
the expression for calculating the linear abnormality degree is as follows:
wherein the content of the first and second substances,
which is indicative of the degree of the linear anomaly,
representing the slope of the curve between the ith element and the last element in the dimension-reduced sequence,
which represents the average slope of the light beam,
is the number of columns of elements in the spectral image;
all the elements of each segmentation region corresponding to the dimensionality reduction sequence are obtained, the slope of a curve between each element of each segmentation region and the last element of the region is obtained, and the linear abnormal degree of each segmentation region is calculated by using an expression for calculating the linear abnormal degree.
105. And setting the element values of all the segmentation areas with the linear abnormal degrees larger than the threshold value as 1 to obtain the road surface abnormal binary image.
And setting a corresponding threshold value Ks, judging that the region is abnormal if the region abnormality degree GD is greater than the threshold value Ks, marking the region as an abnormal region, and obtaining a corresponding abnormal binary image if the element value in the region is 1, or else, obtaining 0.
106. And dividing the road surface abnormal binary image into a plurality of acquisition regions, and controlling a vehicle-mounted rear camera in the corresponding region to acquire the image when the element value in the corresponding acquisition region is 1.
The invention needs to control the acquisition parameters of the rear camera according to the obtained abnormal binary image.
By pairsThe calibration operation of the front camera can obtain the internal reference matrix of the front camera, namely the conversion coefficients of the physical quantities corresponding to the horizontal direction and the longitudinal direction of each pixel in the acquired image, namely
Respectively representing the physical quantities corresponding to the horizontal direction and the vertical direction of a single pixel in the image, and further obtaining the physical size of the acquisition range corresponding to the image:
。
the method for dividing the road surface abnormal binary image into a plurality of acquisition regions comprises the following steps:
acquiring acquisition time according to the number of lines of elements in the road surface abnormal binary image and the running speed of a vehicle where the vehicle-mounted camera is located, and longitudinally dividing the road surface abnormal binary image into a plurality of areas according to the acquisition frequency and the acquisition time of the rear camera;
and horizontally dividing the road surface abnormal binary image into a plurality of acquisition areas according to the number of the rear vehicle-mounted cameras.
Processing the abnormal binary image, and dividing the abnormal binary image into [ n × m ] areas, wherein n is the number of rear cameras, and m is the number of longitudinal judgment areas:
wherein the content of the first and second substances,
the length of the physical range acquired by the front camera,
as the vehicle speed, the distance divided by the speed is the time taken to travel the distance,
is the sampling frequency of the camera, i.e. the number of samples per unit of time, in units of times/second.
Namely the maximum number of times that the camera can collect in the range, so as to uniformly divide the binary image to be processed.
The method for controlling the vehicle-mounted rear camera of the corresponding area to acquire the image of the area comprises the following steps:
acquiring all elements with element values of 1 in a road surface abnormal binary image, and controlling a rear camera to acquire images when the elements with the element values of 1 appear in an acquisition area corresponding to the vehicle-mounted rear camera;
when the element with the element value of 1 does not appear in the acquisition area corresponding to the vehicle-mounted rear camera, the rear camera does not acquire the image.
And obtaining n × m divided regions, and obtaining a matrix with the size of n × m according to the abnormality of each region. Each column is a control sequence corresponding to whether the camera is started or not, the median value of the sequence is 1, which indicates that the position is abnormal, the camera needs to be started for sampling at the moment, otherwise, the camera does not acquire images.
Thus, a road surface image with an abnormality can be acquired. The image information can be sent to a cloud server in a wireless transmission mode, and high-accuracy identification reasoning is carried out by combining a neural network.
107. And inputting the image acquired by the vehicle-mounted rear camera into a neural network, and judging the pavement quality by utilizing the pavement abnormal category output by the neural network.
The neural network can be used for identifying the abnormity with obvious other characteristics for the image collected by the rear camera.
Namely, the cracks on the roads in the image are identified by semantically segmenting the neural network, and the adopted target identification neural network specifically comprises the following contents:
and (3) using a semantic segmentation Unet network, outputting a corresponding semantic segmentation result after data RGB images are processed, extracting image features through convolution and pooling, and then reconstructing the images by adopting deconvolution and anti-pooling operations to obtain semantic segmentation images corresponding to the input images.
A large number of real road pavement images under the view angle of the corresponding automobile data recorder are collected as a data set, to train the neural network.
And obtaining semantic segmentation image label information corresponding to each RGB image by artificially labeling category information in the image, wherein the labeling background pixel category is 0, the road is 1, and the transverse crack is 2. The longitudinal split was 3 and the spider web split was 4.
Because of the classification task, the network uses a cross entropy loss function to supervise the training.
After the neural network training is finished, the images acquired in real time can be sent into the network, corresponding semantic segmentation images are obtained through network reasoning, and then the pavement crack target is obtained through the class labels of the pixels.
After various crack defects are identified, the crack defects can be further visually displayed on a corresponding display, so that the abnormal conditions of the road surface can be more visually checked.
According to the technical means provided by the invention, fourier transformation is carried out on the image, the abnormity is rapidly judged from the frequency domain, and then the rear camera is controlled to collect the abnormal area according to the estimated abnormal area, so that the power consumption of the system is reduced on the whole, the image does not need to be collected all the time, the high-precision abnormal area image is obtained to be identified, the identification precision of a neural network is improved, and the high-precision detection of the pavement quality is realized.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (7)
1. The utility model provides a data intelligent control collection method for traffic engineering road surface quality detects which characterized in that includes:
collecting a pavement gray level image to obtain a frequency spectrum image of the pavement gray level image;
constructing a dimensionality reduction sequence of the spectral image according to the frequency energy mean value of each row of elements in the spectral image; performing curve fitting on the dimensionality reduction sequence of the frequency spectrum image, and calculating the slope of a curve between each element and the last element in the curve;
calculating the size of a segmentation frame according to the sequence number of an element in the dimensionality reduction sequence corresponding to the maximum slope value of the curve between the last elements, and performing sliding segmentation on the frequency spectrum image by using the obtained size of the segmentation frame to obtain a plurality of segmentation areas;
calculating the linear abnormal degree of each segmentation region according to the curve slope between each element and the last element in the dimensionality reduction sequence curve corresponding to each segmentation region;
setting element values of all the segmentation areas with the linear abnormal degrees larger than the threshold value as 1 to obtain a road surface abnormal binary image;
dividing the road surface abnormal binary image into a plurality of acquisition areas, and controlling a vehicle-mounted rear camera of the corresponding area to acquire the image when the element value in the corresponding acquisition area is 1;
and inputting the image acquired by the vehicle-mounted rear camera into a neural network, and judging the pavement quality by utilizing the pavement abnormal category output by the neural network.
2. The method for intelligently regulating and acquiring data for detecting the road surface quality of traffic engineering according to claim 1, wherein the method for constructing the dimensionality reduction sequence of the spectral image according to the frequency energy mean value of each row of elements in the spectral image comprises the following steps:
converting the road surface gray level image with the size of C multiplied by K into a frequency spectrum image by utilizing Fourier transform;
and taking the center of the frequency spectrum as a starting point, acquiring the frequency energy mean value of each row of elements in the right half part of the frequency spectrum image, and constructing a dimension reduction sequence of the frequency spectrum image according to the frequency energy mean value of each row of elements.
3. The method for intelligently regulating and acquiring data for detecting the road surface quality of traffic engineering according to claim 1, wherein the method for calculating the size of the partition frame comprises the following steps:
calculating the slope of the curve between each element and the last element in the curve, and acquiring the average slope of the slopes of all the elements;
obtaining the difference degree of each element according to the square of the difference value between the slope of the curve and the average slope between each element and the last element, and obtaining the serial number value of the element corresponding to the maximum difference degree in the dimension reduction sequence;
and calculating the size of the segmentation frame according to the sequence number value of the element in the dimension reduction sequence and the length of the frequency spectrum image.
4. The method for intelligently regulating and acquiring data for detecting the road surface quality of traffic engineering according to claim 1, wherein the method for performing sliding segmentation on the frequency spectrum image to obtain a plurality of segmented regions comprises the following steps:
obtaining a segmentation direction according to a curve slope and an average slope between an element corresponding to the maximum difference degree and the last element;
wherein the content of the first and second substances,
representing a dividing direction parameter, concave transverse
Convex in the longitudinal direction
，
Representing the slope of the curve between the element corresponding to the greatest degree of difference and the last element,
representing the average slope, and s represents the serial number of the element corresponding to the maximum difference degree in the dimension reduction sequence;
and performing sliding segmentation on the frequency spectrum image in the corresponding segmentation direction by using the segmentation frame to obtain a plurality of segmentation areas.
5. The method for intelligently regulating and acquiring data for detecting the road surface quality of traffic engineering according to claim 1, wherein the method for calculating the linear abnormal degree of each segmented area comprises the following steps:
the expression for calculating the linear abnormality degree is as follows:
wherein the content of the first and second substances,
which is indicative of the degree of the linear anomaly,
representing the slope of the curve between the ith element and the last element in the dimension-reduced sequence,
which represents the average slope of the light beam,
is the number of columns of elements in the spectral image;
and acquiring all corresponding elements of each segmentation region in the dimension reduction sequence, acquiring the slope of a curve between each element in each segmentation region and the last element of the region, and calculating the linear abnormal degree of each segmentation region by using an expression for calculating the linear abnormal degree.
6. The intelligent regulation and control data acquisition method for the road surface quality detection of the traffic engineering according to claim 1, wherein the method for dividing the road surface abnormal binary image into a plurality of acquisition regions comprises the following steps:
acquiring acquisition time according to the number of lines of elements in the road surface abnormal binary image and the running speed of a vehicle where the vehicle-mounted camera is located, and longitudinally dividing the road surface abnormal binary image into a plurality of areas according to the acquisition frequency and the acquisition time of the rear camera;
and horizontally dividing the road surface abnormal binary image into a plurality of acquisition areas according to the number of the rear vehicle-mounted cameras.
7. The method for intelligently regulating and acquiring data for detecting the road surface quality of traffic engineering according to claim 1, wherein the method for controlling the vehicle-mounted rear camera of the corresponding area to acquire the image of the area comprises the following steps:
acquiring all elements with element values of 1 in a road surface abnormal binary image, and controlling a rear camera to acquire images when the elements with the element values of 1 appear in an acquisition area corresponding to the vehicle-mounted rear camera;
when the element with the element value of 1 does not appear in the acquisition area corresponding to the vehicle-mounted rear camera, the rear camera does not acquire the image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211185973.8A CN115272323B (en) | 2022-09-28 | 2022-09-28 | Data intelligent regulation and control acquisition method for traffic engineering pavement quality detection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211185973.8A CN115272323B (en) | 2022-09-28 | 2022-09-28 | Data intelligent regulation and control acquisition method for traffic engineering pavement quality detection |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115272323A true CN115272323A (en) | 2022-11-01 |
| CN115272323B CN115272323B (en) | 2022-12-27 |
Family
ID=83757990
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211185973.8A Active CN115272323B (en) | 2022-09-28 | 2022-09-28 | Data intelligent regulation and control acquisition method for traffic engineering pavement quality detection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115272323B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116630813A (en) * | 2023-07-24 | 2023-08-22 | 青岛奥维特智能科技有限公司 | Highway road surface construction quality intelligent detection system |
| CN117807279A (en) * | 2024-02-29 | 2024-04-02 | 辽宁云也智能信息科技有限公司 | Data retrieval method for highway quality detection |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20200080987A (en) * | 2018-12-27 | 2020-07-07 | 세종대학교산학협력단 | Method and system for analyzing baby behavior |
| CN112712467A (en) * | 2021-01-11 | 2021-04-27 | 郑州科技学院 | Image processing method based on computer vision and color filter array |
| CN114926436A (en) * | 2022-05-20 | 2022-08-19 | 南通沐沐兴晨纺织品有限公司 | Defect detection method for periodic pattern fabric |
-
2022
- 2022-09-28 CN CN202211185973.8A patent/CN115272323B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20200080987A (en) * | 2018-12-27 | 2020-07-07 | 세종대학교산학협력단 | Method and system for analyzing baby behavior |
| CN112712467A (en) * | 2021-01-11 | 2021-04-27 | 郑州科技学院 | Image processing method based on computer vision and color filter array |
| CN114926436A (en) * | 2022-05-20 | 2022-08-19 | 南通沐沐兴晨纺织品有限公司 | Defect detection method for periodic pattern fabric |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116630813A (en) * | 2023-07-24 | 2023-08-22 | 青岛奥维特智能科技有限公司 | Highway road surface construction quality intelligent detection system |
| CN116630813B (en) * | 2023-07-24 | 2023-09-26 | 青岛奥维特智能科技有限公司 | Highway road surface construction quality intelligent detection system |
| CN117807279A (en) * | 2024-02-29 | 2024-04-02 | 辽宁云也智能信息科技有限公司 | Data retrieval method for highway quality detection |
| CN117807279B (en) * | 2024-02-29 | 2024-05-14 | 辽宁云也智能信息科技有限公司 | Data retrieval method for highway quality detection |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115272323B (en) | 2022-12-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115272323B (en) | Data intelligent regulation and control acquisition method for traffic engineering pavement quality detection | |
| CN108711149B (en) | Mineral rock granularity detection method based on image processing | |
| CN114820773B (en) | Silo transport vehicle carriage position detection method based on computer vision | |
| CN109448397B (en) | Group fog monitoring method based on big data | |
| CN117094914B (en) | Smart city road monitoring system based on computer vision | |
| CN104992429A (en) | Mountain crack detection method based on image local reinforcement | |
| CN102393902A (en) | Vehicle color detection method based on H_S two-dimensional histogram and regional color matching | |
| CN112749654A (en) | Deep neural network model construction method, system and device for video fog monitoring | |
| CN102214290B (en) | License plate positioning method and license plate positioning template training method | |
| CN102096826B (en) | Compound classification method for multi-resolution remote sensing image on basis of real likelihood characteristic | |
| CN114882400A (en) | Aggregate detection and classification method based on AI intelligent machine vision technology | |
| CN114998214A (en) | Sampling speed control method and system for cable defect detection | |
| CN115601379A (en) | Surface crack accurate detection technology based on digital image processing | |
| CN107578414B (en) | Method for processing pavement crack image | |
| CN109272484B (en) | Rainfall detection method based on video image | |
| CN112883969B (en) | Rainfall intensity detection method based on convolutional neural network | |
| CN107610094B (en) | Three-dimensional crack detection method based on ellipsoid three-dimensional representation | |
| CN117636268A (en) | Unmanned aerial vehicle aerial natural driving data set construction method oriented to ice and snow environment | |
| CN112614139B (en) | Conveyor belt ore briquette screening method based on depth map | |
| CN110929757B (en) | Method for quickly classifying crack types of asphalt pavement | |
| CN116630311B (en) | Pavement damage identification alarm method for highway administration | |
| CN113139493B (en) | Unmanned road obstacle recognition system | |
| CN114419582B (en) | Abrasion lane line detection method based on phase consistency Hough transformation | |
| CN114066920B (en) | Harvester visual navigation method and system based on improved Segnet image segmentation | |
| CN118096833B (en) | Real-time tracking and detecting method for vehicles in complex road scene |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |