An online monitoring and early warning system for intelligent traffic engineering construction

Abstract

The application relates to the technical field of monitoring and early warning, and discloses an online monitoring and early warning system for intelligent traffic engineering construction, which comprises the following steps: firstly, a single-frame image of a road construction early warning area is acquired, the single-frame image is converted into a developed image of a road plane by using camera calibration parameters, and a long-strip work belt area covering a near-distance sub-area and a far-distance sub-area is calibrated in the image; a two-dimensional Fourier transform is performed on the area to obtain a frequency domain amplitude spectrum; a first sector and a second sector are divided according to a road driving direction; and the direction energy in a medium spatial frequency range is calculated respectively. Further, the road image definition change is analyzed through the energy ratio and the attenuation rate; grade determination is carried out according to a preset threshold value; a visual grade output of a monitoring picture is generated; and intelligent visual monitoring and real-time early warning of the construction area under complex illumination and meteorological conditions are realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of monitoring and early warning, more specifically, it relates to an online monitoring and early warning system for intelligent traffic engineering construction. BACKGROUND

[0002] With the continuous growth of highway construction and maintenance mileage, the construction of occupying the road has become a normalized operation mode. In order to ensure safety, it is usually required to set obvious warning signs and lighting devices within 300 to 500 meters in the direction of the oncoming vehicle. However, in foggy or night environment, the reliability of traditional sound and light warning and camera monitoring decreases significantly. The strong directional light beam of the flashing light in high humidity environment and the water droplets in the air produce strong scattering, which causes the camera image to appear halo, curtain effect and range blur, especially in the construction early warning area (300 to 500 meters) section, the contrast of the picture collapses quickly, and the traditional algorithm cannot stably extract the road boundary or vehicle motion information. Such environmental light interference causes the monitoring system to be visible but not determinable, which directly affects the accuracy of the intelligent early warning chain.

[0003] The flashing light usually flashes at a frequency of 50 to 60 times per minute, and its instantaneous high-brightness pulse forms a light curtain along the driving direction in the foggy air, which optically low-passes the longitudinal texture of the image. This effect is manifested in a single frame image as follows: the spectral energy of the linear structure (marking, median strip edge, etc.) in the far zone of the road decays sharply, while the transverse texture is still partially visible. This non-uniform blur destroys the conventional image analysis model based on spatial gradient or edge detection, resulting in a significant decrease in the recognition rate of the construction monitoring system for the far risk zone. In addition, the automatic exposure compensation mechanism of the night camera will amplify the brightness of the near zone, further suppressing the far zone signal, so that the system cannot accurately perceive the real construction risk environment. SUMMARY

[0004] The present application provides an online monitoring and early warning system for intelligent traffic engineering construction, which solves the technical problems raised in the background art.

[0005] The present application provides an online monitoring and early warning system for intelligent traffic engineering construction, which includes:

[0006] A data acquisition module acquires a single frame image frame of the road construction early warning area;

[0007] An image unwrapping module converts the single frame image frame into an unwrapped image of the road plane based on the camera calibration parameters;

[0008] An image segmentation module labels a long-strip working zone area set along the center strip of the construction lane on the unwrapped image, and the long-strip working zone area covers the near distance sub-area and the far distance sub-area of the road construction early warning area;

[0009] The image transformation module performs two-dimensional Fourier transform on the long strip-shaped working belt area to obtain a two-dimensional Fourier amplitude spectrum;

[0010] The sector extraction module divides the first sector and the second sector according to the driving direction of the road, wherein the first sector is consistent with the driving direction of the road, and the second sector is orthogonal to the driving direction of the road.

[0011] The energy calculation module calculates the energy values of the first sector and the second sector in a preset medium spatial frequency range to obtain the first direction energy and the second direction energy.

[0012] The attenuation calculation module determines the direction energy attenuation rate based on the first direction energy and the second direction energy.

[0013] The risk early warning module performs grade determination on the single-frame direction energy ratio and the direction energy attenuation rate according to a preset threshold interval to generate a visual level output result of the monitoring picture.

[0014] The beneficial effects of the present application include: by introducing the image unwrapping and two-dimensional Fourier spectrum analysis technology based on the camera calibration parameters, the stable online monitoring and visual early warning of the road construction early warning zone in the low-visibility environment such as foggy day and night are realized. The present application uses the frequency domain direction energy analysis method to effectively distinguish the texture attenuation characteristics of the driving direction and the orthogonal direction, and overcomes the recognition failure problem of the traditional image algorithm under strong scattered light and flash interference. By constructing the double-index determination model of the direction energy ratio and the attenuation rate, the system can output the visual level of the construction monitoring picture in real time, and realize the intelligent jump from visible but undeterminable to determinable and warning. Compared with the existing sound-light alarm or ordinary video monitoring scheme, the present application significantly improves the image determinability and early warning reliability under complex weather conditions, has the outstanding advantages of no need of additional hardware investment, strong algorithm robustness, wide application range, etc. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 is a module diagram of an online monitoring and early warning system for intelligent traffic engineering construction of the present application. DETAILED DESCRIPTION

[0016] The subject matter described herein will now be discussed with reference to example implementations. It should be understood that the discussion of these implementations is merely meant to provide a better understanding of the subject matter described herein and can be changed in function and arrangement without departing from the scope of the present description. Various processes or components can be omitted, substituted, or added according to desired implementations. Additionally, features described with respect to some examples can be combined in other examples.

[0017] As Figure 1As shown, an online monitoring and early warning system for intelligent traffic engineering construction comprises:

[0018] A data acquisition module acquires a single-frame image frame of a road construction early warning area;

[0019] An image unwrapping module converts the single-frame image frame into an unwrapped image of a road plane based on camera calibration parameters;

[0020] An image segmentation module labels a long-strip work zone area arranged along a construction lane center strip on the unwrapped image, and the long-strip work zone area covers a near-distance sub-area and a far-distance sub-area of the road construction early warning area;

[0021] An image transformation module performs two-dimensional Fourier transform on the long-strip work zone area to obtain a two-dimensional Fourier amplitude spectrum;

[0022] A sector extraction module divides a first sector and a second sector according to a road driving direction; wherein the first sector is consistent with the road driving direction, and the second sector is orthogonal to the road driving direction;

[0023] An energy calculation module calculates energy values of the first sector and the second sector in a preset medium spatial frequency range to obtain a first direction energy and a second direction energy;

[0024] An attenuation calculation module determines a direction energy attenuation rate based on the first direction energy and the second direction energy;

[0025] A risk early warning module performs level determination on a single-frame direction energy ratio and the direction energy attenuation rate according to a preset threshold interval to generate a visual level output result of a monitoring picture.

[0026] In an embodiment of the present application, converting the single-frame image frame into the unwrapped image of the road plane based on the camera calibration parameters comprises:

[0027] Determining a camera intrinsic parameter matrix, a camera extrinsic parameter rotation matrix, a camera extrinsic parameter translation vector, a road ground normal vector, and a directed distance from the road ground to a camera coordinate system;

[0028] According to the determined camera intrinsic parameter matrix, the camera extrinsic parameter rotation matrix, the camera extrinsic parameter translation vector, the road ground normal vector, and the directed distance from the road ground to the camera coordinate system, a homography matrix is calculated;

[0029] Homogeneous coordinates of each image pixel point in the single-frame image frame are obtained, and the homogeneous coordinates of each image pixel point are mapped by the homography matrix to obtain homogeneous coordinates of corresponding points in the unwrapped image of the road plane;

[0030] According to the inverse matrix of the homographic matrix, inverse mapping is performed on the homogeneous coordinates of each corresponding point in the road plane unwrapping image to obtain the homogeneous coordinates of each corresponding point in the single frame image frame;

[0031] The homogeneous coordinates of each corresponding point in the road plane unwrapping image are converted into non-homogeneous coordinates; wherein the non-homogeneous coordinates are obtained by taking the quotient of the first component and the second component of the corresponding homogeneous coordinates and the third component;

[0032] Based on the non-homogeneous coordinates of each corresponding point, the corresponding pixel value in the single frame image frame is obtained, the road plane unwrapping image is resampled, and the road plane unwrapping image is formed.

[0033] The homographic matrix is calculated ; wherein, represents the camera intrinsic parameter matrix, represents the camera extrinsic parameter translation vector, represents the directed distance from the road ground to the camera coordinate system, represents the road ground normal vector, represents the camera extrinsic parameter rotation matrix.

[0034] It should be noted that the homogeneous coordinates of the pixel point in the single frame image frame are obtained , the homogeneous coordinates are three-dimensional vectors; mapping is performed on the homogeneous coordinates by the homographic matrix , and the mapping formula is: , to obtain the homogeneous coordinates of the corresponding point in the road plane unwrapping image ; wherein, is a homogeneous scale factor, and the homogeneous coordinates are three-dimensional vectors; the inverse matrix of the homographic matrix is calculated, and the road plane unwrapping image is resampled by . Non-homogeneous processing is performed on the homogeneous coordinates to obtain the non-homogeneous coordinates of the corresponding point in the road plane unwrapping image , and the non-homogeneous coordinates , wherein , , are the first, second and third components of the homogeneous coordinates , respectively, and the road plane unwrapping image is generated according to the non-homogeneous coordinates .

[0035] Camera intrinsic parameter matrix: describes the inherent parameters of the camera itself optical characteristics (such as lens focal length, pixel size) and imaging plane, which determines how the light is imaged inside the camera.

[0036] Camera extrinsic rotation matrix: Reflects the camera's attitude angle relative to the road plane (e.g., whether the camera is tilted or deflected), used to correct the image offset caused by the camera's angle.

[0037] Camera extrinsic translation vector: Describes the camera's position coordinates in three-dimensional space relative to the road plane (e.g., the camera's height from the road surface, the horizontal distance from the construction area), used to locate the camera's spatial position relative to the road.

[0038] Road surface normal vector: Represents the spatial orientation of the road plane (e.g., whether the road surface is level or has a slight slope), ensuring that subsequent transformations conform to the real road surface's spatial form.

[0039] Road surface to camera coordinate system's directed distance: Quantifies the vertical / horizontal distance between the road plane and the camera coordinate origin, providing a scale reference for coordinate conversion.

[0040] Homography matrix is used to associate image pixel coordinates with road plane spatial coordinates. Thus, the near-large and far-small distortion caused by perspective in the image is eliminated, establishing a one-to-one correspondence between pixel points and real space points on the road plane.

[0041] Obtain pixel homogeneous coordinates: Convert the two-dimensional coordinates of each pixel in a single frame image (e.g., horizontal and vertical pixel positions) into three-dimensional homogeneous coordinates (add a scale component). This coordinate form can more accurately handle scale issues in projection transformation and avoid calculation errors.

[0042] Map to get the unfolded image homogeneous coordinates: Use the homography matrix to operate on the homogeneous coordinates of each pixel to obtain the corresponding homogeneous coordinates of the pixel in the road plane unfolded image. This is equivalent to projecting the pixel points in the image under the perspective view onto the real road plane, initially determining their positions in the unfolded image.

[0043] Divide the first two components of the homogeneous coordinates (corresponding to horizontal and vertical coordinates, respectively) by the third component (scale component) to obtain two-dimensional non-homogeneous coordinates. Non-homogeneous coordinates can directly correspond to the actual pixel positions of the original image (e.g., horizontal Xth pixel, vertical Yth pixel), achieving dimension reduction and adaptation of coordinates.

[0044] Based on non-homogeneous coordinates, extract the pixel values (e.g., grayscale information) at the corresponding positions from the original single frame image, and fill these pixel values into the corresponding positions of the unfolded image. After filling the pixel values for all points in the unfolded image, the road plane unfolded image is generated. This image has no perspective distortion, and the pixel positions are consistent with the real spatial scale of the road (e.g., the distance between adjacent pixels in the unfolded image corresponds to the actual distance in meters on the road surface).

[0045] In one embodiment of the present application, a long-strip work zone region is marked on the expanded image along the center strip of the construction lane, the long-strip work zone region covers the near-distance sub-region and the far-distance sub-region of the road construction early warning area, including:

[0046] The center point coordinates of the near-distance sub-region, the center point coordinates of the far-distance sub-region, the lane width, the meter per pixel scale, the unit matrix, the lower boundary distance of the near-distance sub-region, the upper boundary distance of the near-distance sub-region, the lower boundary distance of the far-distance sub-region, and the upper boundary distance of the far-distance sub-region are set;

[0047] According to the center point coordinates of the near-distance sub-region and the center point coordinates of the far-distance sub-region, the construction lane direction vector and the center strip length are calculated, the construction lane direction vector is used to determine the longitudinal direction of the long-strip work zone region, and the center strip length is used to limit the longitudinal range of the long-strip work zone region;

[0048] For each point in the expanded image, according to the construction lane direction vector and the difference between the point and the center point coordinates of the near-distance sub-region, a longitudinal projection scalar is calculated, which is used to measure the longitudinal position of the point on the construction lane center strip;

[0049] According to the unit matrix, the construction lane direction vector, and the difference between the point and the center point coordinates of the near-distance sub-region, a transverse distance scalar is calculated, which is used to measure the transverse distance of the point to the construction lane center strip;

[0050] The long-strip work zone region is defined as the region in the expanded image where the longitudinal projection scalar is between 0 and the center strip length and the transverse distance scalar does not exceed half of the lane width;

[0051] The lower boundary distance and the upper boundary distance of the near-distance sub-region are divided by the meter per pixel scale to obtain the longitudinal pixel boundary of the near-distance sub-region;

[0052] The part of the long-strip work zone region where the longitudinal projection scalar is between the longitudinal pixel boundaries of the near-distance sub-region is the near-distance sub-region;

[0053] The lower boundary distance and the upper boundary distance of the far-distance sub-region are divided by the meter per pixel scale to obtain the longitudinal pixel boundary of the far-distance sub-region;

[0054] The part of the long-strip work zone region where the longitudinal projection scalar is between the longitudinal pixel boundaries of the far-distance sub-region is the far-distance sub-region.

[0055] The construction lane direction vector parameter is calculated ; wherein, represents the center point coordinates vector of the far-distance sub-region, represents the center point coordinates vector of the near-distance sub-region, denotes the module length of

[0056] Calculate the center strip length parameter .

[0057] Calculate the longitudinal projection scalar parameter for a point in the unwrapped image where is the transpose of the construction lane direction vector parameter .

[0058] Calculate the transverse distance scalar parameter for a point in the unwrapped image where is the identity matrix parameter minus the product of the construction lane direction vector parameter and .

[0059] Define the long strip work zone area ; where denotes the lane width parameter.

[0060] Calculate the longitudinal pixel boundary parameters and of the near distance sub-area. Wherein, denotes the meter per pixel scale parameter, denotes the near distance sub-area lower boundary distance parameter, denotes the near distance sub-area upper boundary distance parameter.

[0061] Define the near distance sub-area .

[0062] Calculate the longitudinal pixel boundary parameters and of the far distance sub-area. Wherein, denotes the far distance sub-area lower boundary distance parameter, denotes the far distance sub-area upper boundary distance parameter.

[0063] Near / far distance sub-area center point coordinates: locate the core positions of near and far sub-areas respectively (such as the near distance center point corresponds to the midpoint of the 50-80 meter interval of the road surface, and the far distance center point corresponds to the midpoint of the 300-500 meter interval).

[0064] Lane width: adopt the standard lane width (3.5-3.75 meters) to limit the transverse range of the work zone.

[0065] Meter per pixel scale: clearly define the number of meters corresponding to one pixel in the unwrapped image on the real road surface, realizing the scale conversion between image pixels and actual road distance.

[0066] Identity matrix: used for calculating the lateral distance of each point to the lane center strip, to correct the direction deviation in coordinate calculation.

[0067] Near / far sub-region lower / upper boundary distance: define the longitudinal start / end distance of near / far sub-region on real road surface (e.g. near distance lower boundary 50m, upper boundary 80m), to provide actual distance basis for sub-region division.

[0068] Construction lane direction vector: calculated by the difference between the center point coordinates of far sub-region and near sub-region, combined with the length of the difference (straight-line distance between two points). This vector can reflect the longitudinal extension direction of the lane.

[0069] Center strip length: the straight-line distance between the center points of near and far sub-regions, used to limit the longitudinal total range of long strip working zone, to ensure that the working zone covers the complete construction warning area from near sub-region to far sub-region.

[0070] For each pixel point in the expanded image, calculate the longitudinal projection scalar by the difference between the construction lane direction vector and the coordinate of the point to the center point of near sub-region. The role of this scalar is to quantify the position of each point on the longitudinal axis of the lane center strip. If the scalar is between 0 and the center strip length, it means that the longitudinal position of the point is between the near and far sub-regions, meeting the longitudinal requirements of the working zone.

[0071] Calculate the lateral distance scalar by the difference between the identity matrix construction lane direction vector and the coordinate of the point to the center point of near sub-region. The role of this scalar is to quantify the lateral distance of each point to the lane center strip. If the scalar does not exceed half of the lane width, it means that the lateral position of the point is within the lane range, meeting the lateral requirements of the working zone.

[0072] Define all pixel points in the expanded image with longitudinal projection scalar between 0 and center strip length and lateral distance scalar not exceeding half of the lane width as the long strip working zone area. This area not only covers the complete warning area along the longitudinal direction of the lane, but also is limited within a single lane along the lateral direction, matching the target range of construction monitoring.

[0073] Divide the near distance sub-region lower boundary distance and near distance sub-region upper boundary distance (both are real road surface distances) by the pixel scale of 1 meter, to get the corresponding longitudinal pixel boundaries. This step is to convert the start / end distance of near sub-region on real road surface into pixel coordinates of the expanded image.

[0074] In the long strip working zone area, select the pixel points with longitudinal projection scalar between the near distance sub-region longitudinal pixel boundaries, to form the near distance sub-region. This area corresponds to the 50-80m interval on real road surface.

[0075] The distance between the lower boundary of the long-distance sub-region and the upper boundary of the long-distance sub-region (the real road distance, such as 300 meters, 500 meters) is divided by the meter per pixel scale respectively to obtain the corresponding longitudinal pixel boundary, and the conversion from the real distance to the image pixel is completed.

[0076] In the long-strip-shaped working belt region, the pixel points with the longitudinal projection scalar between the longitudinal pixel boundaries of the long-distance sub-region are screened out to form the long-distance sub-region. The region corresponds to the 300-500 meter construction early warning area of the real road.

[0077] In an embodiment of the present application, a two-dimensional Fourier transform is performed on the long-strip-shaped working belt region to obtain a two-dimensional Fourier amplitude spectrum, including:

[0078] The total number of horizontal pixels and the total number of longitudinal pixels of the long-strip-shaped working belt region are determined;

[0079] The long-strip-shaped working belt region is represented as a gray scale function, which takes the horizontal pixel position and the longitudinal pixel position in the long-strip-shaped working belt region as variables;

[0080] The horizontal frequency count, the longitudinal frequency count and the imaginary unit are set;

[0081] Based on the total number of horizontal pixels, the total number of longitudinal pixels, the gray scale function, the horizontal frequency count, the longitudinal frequency count and the imaginary unit, a two-dimensional discrete Fourier transform is performed on the long-strip-shaped working belt region to obtain a complex spectrum of the two-dimensional discrete Fourier transform;

[0082] The absolute value of the complex spectrum of the two-dimensional discrete Fourier transform is calculated to obtain a two-dimensional Fourier amplitude spectrum.

[0083] The complex spectrum of the two-dimensional discrete Fourier transform is calculated , and the calculation formula is as follows:

[0084] ; wherein, is the horizontal coordinate of the pixel in the long-strip-shaped working belt region, is the longitudinal coordinate of the pixel in the long-strip-shaped working belt region, is the gray scale function, is the imaginary unit, is the total number of horizontal pixels, is the total number of longitudinal pixels, is the longitudinal frequency count, is the horizontal frequency count.

[0085] The two-dimensional Fourier amplitude spectrum is calculated ; wherein, represents the modulus.

[0086] The total number of horizontal pixels (the number of pixels contained in the horizontal direction) and the total number of vertical pixels (the number of pixels contained in the vertical direction) of the long strip-shaped working zone are counted. The purpose of this operation is to determine the spatial dimension range of the Fourier transform.

[0087] The long strip-shaped working zone is represented as a gray function, the input of which is the horizontal pixel position and the vertical pixel position of the pixels in the zone, and the output is the gray value (such as 0-255 gray level) of the corresponding position pixel. Thus, the two-dimensional pixel array of the image is converted into a mathematical function, that is, the texture characteristics of the road surface (such as the gray difference between the marking and the road surface, the gray fluctuation of the joint) will be reflected in the value change of the function.

[0088] The horizontal frequency count, the vertical frequency count and the imaginary unit are set:

[0089] Horizontal / vertical frequency count: corresponding to the frequency component serial number in the horizontal direction and the vertical direction respectively, each count represents a specific frequency (such as low count corresponding to low frequency, high count corresponding to high frequency).

[0090] Imaginary unit: Fourier transform is a mathematical tool for processing periodic signals, and the imaginary unit is a necessary element of the transform operation, which is used to describe the phase information of the signal.

[0091] The two-dimensional discrete Fourier transform is performed on the long strip-shaped working zone. The purpose of this transform is to decompose the change of the gray value with the pixel position (i.e. the road surface texture) in the spatial domain into a combination of sine / cosine signals of different frequencies, and finally output a two-dimensional discrete Fourier complex spectrum. Each element of the complex spectrum contains two parts of information, i.e. the amplitude (corresponding to the intensity of a certain frequency component, which is positively correlated with the clarity of the texture) and the phase (corresponding to the position information of a certain frequency component).

[0092] For each element in the two-dimensional discrete Fourier complex spectrum, the absolute value (mathematically referred to as the modulus) is calculated, and finally a two-dimensional Fourier amplitude spectrum is obtained. The purpose of this step is to eliminate the phase information and retain the energy information, that is, the value at each position in the amplitude spectrum represents the energy intensity of the corresponding frequency component.

[0093] In one embodiment of the present application, the first sector and the second sector are divided according to the driving direction of the road; wherein the first sector is consistent with the driving direction of the road, and the second sector is orthogonal to the driving direction of the road, comprising:

[0094] The angle half-width is determined, and the unit of the angle half-width is radian; the driving direction angle of the road is calculated according to the horizontal component and the vertical component of the construction lane direction vector.

[0095] The road driving direction unit vector is constructed based on the cosine value and the sine value of the road driving direction angle, and the transverse component of the road driving direction unit vector is the cosine value of the road driving direction angle, and the longitudinal component is the sine value of the road driving direction angle.

[0096] The road driving direction orthogonal unit vector is constructed based on the sine value and the cosine value of the road driving direction angle, and the transverse component of the road driving direction orthogonal unit vector is the negative value of the sine value of the road driving direction angle, and the longitudinal component is the cosine value of the road driving direction angle.

[0097] For each frequency point in the two-dimensional Fourier amplitude spectrum, the frequency vector of the frequency point is calculated, the transverse component of the frequency vector is the transverse frequency count of the frequency point minus half of the total number of transverse pixels, and the longitudinal component of the frequency vector is the longitudinal frequency count of the frequency point minus half of the total number of longitudinal pixels.

[0098] For each frequency vector, when the length of the frequency vector is not zero, the frequency vector is normalized to obtain the frequency direction unit vector of the frequency point.

[0099] All frequency points in the two-dimensional Fourier amplitude spectrum that satisfy a first preset condition are grouped into a first sector, wherein the first preset condition is that the dot product of the frequency direction unit vector and the road driving direction unit vector is not less than the cosine value of the angle half width.

[0100] All frequency points in the two-dimensional Fourier amplitude spectrum that satisfy a second preset condition are grouped into a second sector, wherein the second preset condition is that the dot product of the frequency direction unit vector and the road driving direction orthogonal unit vector is not less than the cosine value of the angle half width.

[0101] The road driving direction angle parameter is calculated ; wherein is an arctangent function, the longitudinal component of the construction lane direction vector parameter is , the transverse component of the opposite side is the adjacent side angle.

[0102] The frequency vector parameter is calculated .

[0103] The frequency direction unit vector parameter is calculated ; wherein is the modulus of the frequency vector parameter .

[0104] The road driving direction unit vector parameter and the orthogonal unit vector parameter are calculated:

[0105] , ;in, Road driving direction angle parameters cosine value, Road driving direction angle parameters The sine value.

[0106] Will satisfy and frequency points The set is defined as the first sector, denoted as ,in, Angular half-width parameter The cosine value will satisfy... and frequency points The set is defined as the second sector, denoted as .

[0107] Half-width angle: Set an angle value in radians (e.g., a radian value corresponding to 15°) to limit the directional range of the sector. The smaller the half-width, the narrower the sector, and the more concentrated the selected frequency points are, avoiding the mixing of frequency components with irrelevant directions.

[0108] Road driving direction angle: Calculated using the arctangent function, with the longitudinal component of the construction lane direction vector as the opposite side and the lateral component as the adjacent side. This angle accurately reflects the actual driving direction of the road.

[0109] Road driving direction unit vector: This vector is constructed using the road driving direction angle, taking the cosine value as the lateral component and the sine value as the longitudinal component. The length of this vector is 1, retaining only the driving direction information, and serves as the direction reference for the first sector.

[0110] Orthogonal unit vector for road driving direction: Also based on the driving direction angle, the negative value of its sine is taken as the lateral component and the cosine value as the longitudinal component to construct this vector. This vector is perpendicular to the unit vector of driving direction (with an angle of 90 degrees), and only the directional information orthogonal to the driving direction is retained as the directional reference for the second sector.

[0111] For each frequency point within the amplitude spectrum, calculate its horizontal and vertical frequency vector components: the horizontal component is the horizontal frequency count of that frequency point minus half the total number of horizontal pixels in the long strip working band region; the vertical component is the vertical frequency count of that frequency point minus half the total number of vertical pixels. This operation shifts the frequency center of the frequency spectrum to the geometric center of the amplitude spectrum.

[0112] For each frequency vector, if its length (modulus) is not zero, the vector is divided by its length to complete the normalization process, and the frequency direction unit vector is obtained. After normalization, the vector only retains the direction information of the frequency point, and eliminates the interference of the length (i.e. the frequency intensity).

[0113] First sector division: filter out the frequency points in the amplitude spectrum that meet two conditions: one is that the length of the frequency vector is not zero (exclude meaningless frequency points), and the other is that the dot product of the frequency direction unit vector and the unit vector of the road driving direction is not less than the cosine value of the angle half-width. The greater the dot product, the closer the direction of the two vectors, so the direction of these frequency points is consistent with the driving direction, forming the first sector.

[0114] Second sector division: filter out the frequency points that meet two conditions: one is that the length of the frequency vector is not zero, and the other is that the dot product of the frequency direction unit vector and the orthogonal unit vector of the road driving direction is not less than the cosine value of the angle half-width. The direction of these frequency points is orthogonal to the driving direction, forming the second sector.

[0115] In an embodiment of the present application, the energy values of the first sector and the second sector in the preset medium spatial frequency range are calculated to obtain the first direction energy and the second direction energy, comprising:

[0116] Set the lower limit of the spatial frequency radius and the upper limit of the spatial frequency radius;

[0117] For each frequency point in the two-dimensional Fourier amplitude spectrum, calculate the spatial frequency radius of the frequency point, which is the distance from the frequency point to the frequency center, and the horizontal coordinate of the frequency center is half of the total number of horizontal pixels, and the vertical coordinate is half of the total number of vertical pixels;

[0118] Define the medium spatial frequency range as a set consisting of all frequency points in the two-dimensional Fourier amplitude spectrum whose spatial frequency radius is not less than the lower limit of the spatial frequency radius and not greater than the upper limit of the spatial frequency radius;

[0119] Calculate the intersection of the first sector and the medium spatial frequency range to obtain the first direction frequency band;

[0120] Calculate the intersection of the second sector and the medium spatial frequency range to obtain the second direction frequency band;

[0121] Calculate the sum of the squares of the two-dimensional Fourier amplitude spectrum of all frequency points in the first direction frequency band to obtain the first direction energy;

[0122] Calculate the sum of the squares of the two-dimensional Fourier amplitude spectrum of all frequency points in the second direction frequency band to obtain the second direction energy.

[0123] Calculate the spatial frequency radius parameter .

[0124] Defining a medium spatial frequency range set parameter ; wherein, represents a spatial frequency radius lower limit parameter, represents a spatial frequency radius upper limit parameter.

[0125] Calculating a first direction band set parameter , calculating a second direction band set parameter .

[0126] Calculating a first direction energy ;

[0127] Calculating a second direction energy .

[0128] Two parameters, a spatial frequency radius lower limit and a spatial frequency radius upper limit, are preset. The two parameters serve to circumscribe the medium spatial frequency range: low frequencies (radius less than the lower limit) are susceptible to image overall brightness, fog curtain effect interference, and cannot reflect texture details; high frequencies (radius greater than the upper limit) are mostly sensor noise or image sharpening artifacts, and also do not represent effective road texture, so the medium frequency interval that can truly reflect the road texture needs to be screened out through the upper and lower limits.

[0129] All frequency points in the two-dimensional Fourier amplitude spectrum with a spatial frequency radius not less than the lower limit and not greater than the upper limit are composed into a medium spatial frequency range.

[0130] First direction band: calculate the intersection of the first sector (a frequency point set consistent with the driving direction) and the medium spatial frequency range (a frequency point set screened out). That is, the frequency point set that simultaneously satisfies the direction consistent with the driving and the medium spatial frequency, focusing on the effective texture frequency in the driving direction;

[0131] Second direction band: calculate the intersection of the second sector (a frequency point set orthogonal to the driving direction) and the medium spatial frequency range. That is, the frequency point set that simultaneously satisfies the direction orthogonal to the driving and the medium spatial frequency, focusing on the effective texture frequency in the orthogonal direction.

[0132] First direction energy: square the two-dimensional Fourier amplitude spectrum values of all frequency points in the first direction band respectively (the square of the amplitude corresponds to the energy in the frequency domain), and then sum all the squared values. The sum obtained is the total energy in the medium frequency range in the driving direction, reflecting the clarity of the driving direction texture (the higher the energy, the clearer the texture);

[0133] Second direction energy: square the two-dimensional Fourier amplitude spectrum values of all frequency points in the second direction band and then sum them, obtaining the total energy in the medium frequency range in the orthogonal direction, reflecting the clarity of the orthogonal direction texture.

[0134] In an embodiment of the present application, the direction energy attenuation rate is determined based on the first direction energy and the second direction energy, comprising:

[0135] A stability constant is set, and the stability constant is a positive number;

[0136] A direction energy ratio is calculated, and the direction energy ratio is the sum of the first direction energy and the stability constant divided by the sum of the second direction energy and the stability constant;

[0137] The direction energy ratio is taken as a natural logarithm;

[0138] The negative value of the natural logarithm is taken to obtain the direction energy attenuation rate.

[0139] A direction energy ratio parameter is calculated ; wherein, The stability constant parameter is represented.

[0140] The direction energy attenuation rate is calculated .

[0141] The sum of the first direction energy and the stability constant is taken as a numerator, and the sum of the second direction energy and the stability constant is taken as a denominator, and the two are divided to obtain the direction energy ratio. Through the processing of adding the stability constant, the calculation error in an extreme case (such as the second direction energy close to zero) is avoided; at the same time, the ratio directly reflects the relative relationship between the two direction energies. If the fog flash scattering causes the longitudinal (driving direction) texture to attenuate, the first direction energy will be less than the second direction energy, and the ratio will be less than 1; when there is no attenuation, the ratio is close to 1.

[0142] The natural logarithm operation is performed on the direction energy ratio. Because the texture energy attenuation caused by the fog flash scattering approximately conforms to the exponential attenuation rule, the natural logarithm can convert this nonlinear exponential relationship into a linear relationship, so that the subsequent calculation of the attenuation rate is more in line with the actual process, and the attenuation degree is easy to quantify.

[0143] The negative value of the natural logarithm result is taken to finally obtain the direction energy attenuation rate. Thus, the case that the logarithm result is negative when the ratio is less than 1 is converted into a positive attenuation rate, that is, the greater the attenuation rate, the more serious the attenuation of the first direction energy (driving direction) relative to the second direction energy (orthogonal direction), and the stronger the fog flash interference; when the attenuation rate is close to zero, it indicates that the difference between the two direction energies is small, and there is no obvious attenuation.

[0144] In an embodiment of the present application, the single-frame direction energy ratio and the direction energy attenuation rate are determined according to a preset threshold interval, a visual level output result of a monitoring picture is generated, comprising:

[0145] The single-frame direction energy ratio and the single-frame direction energy attenuation rate are obtained;

[0146] Set the ratio threshold high, the ratio threshold medium, the ratio threshold low, and the ratio threshold high is greater than the ratio threshold medium is greater than the ratio threshold low is greater than zero;

[0147] Set the decay rate threshold low, the decay rate threshold medium, the decay rate threshold high, and the decay rate threshold low is less than or equal to the decay rate threshold medium is less than the decay rate threshold high;

[0148] Define four visual level values, and the first visual level value is less than the second visual level value is less than the third visual level value is less than the fourth visual level value;

[0149] When the directional energy ratio is greater than or equal to the ratio threshold high, the first visual level value is corresponding;

[0150] When the directional energy ratio is greater than or equal to the ratio threshold medium and less than the ratio threshold high, the second visual level value is corresponding;

[0151] When the directional energy ratio is greater than or equal to the ratio threshold low and less than the ratio threshold medium, the third visual level value is corresponding;

[0152] When the directional energy ratio is less than the ratio threshold low, the fourth visual level value is corresponding;

[0153] When the directional energy decay rate is less than the decay rate threshold low, the first visual level value is corresponding;

[0154] When the directional energy decay rate is greater than or equal to the decay rate threshold low and less than the decay rate threshold medium, the second visual level value is corresponding;

[0155] When the directional energy decay rate is greater than or equal to the decay rate threshold medium and less than the decay rate threshold high, the third visual level value is corresponding;

[0156] When the directional energy decay rate is greater than or equal to the decay rate threshold high, the fourth visual level value is corresponding;

[0157] Take the larger value of the visual level value obtained by the directional energy ratio and the visual level value obtained by the directional energy decay rate as the visual level output result of the monitoring picture.

[0158] Directional energy ratio: derived from the medium spatial frequency energy ratio of the first sector in the driving direction and the second sector in the orthogonal direction, directly reflects the energy difference of longitudinal texture (such as marking line, joint) and transverse texture (the smaller the ratio, the more serious the attenuation of longitudinal texture due to fog flash scattering);

[0159] Directional energy decay rate: is a quantitative index after logarithmic conversion of the ratio, which converts the relative energy difference into an attenuation intensity related to distance (the greater the decay rate, the stronger the fog flash interference in the 300-500 meter construction early warning area).

[0160] The direction energy ratio and the direction energy decay rate jointly constitute the dual-dimension judgment of the system, avoiding the interference of extreme noise (such as local pixel anomalies) to a single index, ensuring the robustness of the judgment basis, and meeting the needs of stable monitoring under complex weather conditions.

[0161] The system sets the ratio threshold (high / medium / low) and the decay rate threshold (low / medium / high), and strictly follows the hierarchy of ratio threshold high>medium>low and decay rate threshold low≤medium<high. The design basis is the actual interference scene of the construction early warning area (300-500 meters):

[0162] For the direction energy ratio: the higher the threshold, the closer the energy of the driving direction and the orthogonal direction (such as the ratio≥high threshold, the longitudinal texture has no obvious decay), corresponding to weak fog flash interference; the lower the threshold, the more serious the longitudinal energy decay (such as the ratio<low threshold, the longitudinal texture is almost invisible due to thick fog / strong flash), corresponding to strong fog flash interference;

[0163] For the direction energy decay rate: the lower the threshold, the smaller the decay intensity of the longitudinal texture (such as the decay rate<low threshold, the influence of fog flash on the texture can be ignored); the higher the threshold, the greater the decay intensity (such as the decay rate≥high threshold, the texture in the 300-500 meter area is seriously blurred), which is positively correlated with the interference intensity.

[0164] The specific value of the threshold (such as the high threshold of the ratio is set to 0.7 and the high threshold of the decay rate is set to 8×10⁻³1 / m) is not subjective, but is calibrated in combination with the hardware parameters of the system flash LED device and the roadside camera:

[0165] Referring to the 50-60 times / minute flashing frequency of the flash LED (MUTCD standard), the critical values of the ratio and the decay rate are determined by measuring the longitudinal texture decay law under different fog concentrations;

[0166] In combination with the 300-500 meter visibility coverage of the camera, it is ensured that the threshold can accurately reflect whether the far area (300-500 meters) can be judged. For example, when the decay rate≥high threshold, the texture in the far area cannot support vehicle recognition, and the highest level of warning needs to be triggered, which meets the core functional positioning of the system early warning.

[0167] Four visible level values are defined in the operation, and the level values increase, which converts the abstract interference intensity into concrete warning instructions:

[0168] The level values from small to large correspond to the interference intensity from weak to strong (such as the first level value corresponds to no interference, and the fourth level value corresponds to strong interference);

[0169] This level is not only a picture display level, but also a key intermediate signal of the system calculation-transmission-link, which will be transmitted to the intelligent control integrated terminal in the future, and is related to the warning action:

[0170] Level 1 (No Interference): No need for enhanced warnings, just maintain basic audio-visual cues;

[0171] Level 2 (weak interference): Moderately increase the brightness of the flashing lights and push navigation prompts to passing vehicles, reminding them to slow down due to construction ahead;

[0172] Level 3 (Medium Interference): Activate high-frequency audible and visual alarms and limit the speed of passing vehicles (e.g., reduce to below 60km / h).

[0173] Level 4 (Strong Interference): Triggers the highest level warning, coordinating with traffic management departments to guide vehicles to detour and avoid accidents caused by invisible textures in distant areas.

[0174] Hierarchical mapping of directional energy ratios: focusing on texture energy differences;

[0175] When the ratio is greater than or equal to the high threshold: the energy in the driving direction and the orthogonal direction are close, and the longitudinal texture is clear (such as a clear night without fog) → corresponding level 1 (no interference).

[0176] When the ratio is greater than or equal to the medium threshold and less than the high threshold: slight attenuation of vertical texture (such as fog) → corresponding level 2 (weak interference);

[0177] When the ratio is greater than or equal to the low threshold and less than the medium threshold: moderate attenuation of vertical texture (e.g., medium fog) → corresponding level 3 (medium interference);

[0178] When the ratio is less than the low threshold: vertical texture is severely attenuated (e.g., dense fog) → corresponding to level 4 (strong interference). This mapping is directly based on texture visibility, which aligns with the monitoring logic of determining whether distant areas are identifiable through texture.

[0179] Mapping of directional energy attenuation rate levels: correlation between focusing attenuation intensity and distance;

[0180] When the attenuation rate is less than the low threshold: longitudinal attenuation is weak, and texture can be judged in the distant area of ​​300-500 meters → corresponding level 1;

[0181] When the attenuation rate is ≥ low and < medium threshold: the distant texture is slightly blurred → corresponding level 2;

[0182] When the attenuation rate is ≥ medium and < high threshold: moderate blurring of distant texture → corresponding level 3;

[0183] When the attenuation rate is greater than or equal to the high threshold: the texture in the far-field area is severely blurred, and vehicles have difficulty identifying the construction area in advance → corresponding to level 4. This mapping additionally associates the spatial scale of the 300-500 meter advance zone (the dimension of the attenuation rate is 1 / meter) to ensure that the level determination is highly consistent with the needs of early warning, that is, to avoid the warning lag caused by only focusing on energy differences and ignoring the actual visibility in the far-field area.

[0184] Scene example 1: If the direction energy ratio corresponds to level 2 (weak interference), but the direction energy decay rate corresponds to level 3 (medium interference), it means that the energy difference seems to be slight, but the far zone decay has reached a medium degree. At this time, level 3 is taken, avoiding insufficient early warning due to misjudgment of a single indicator;

[0185] Scene example 2: If the ratio corresponds to level 3 (medium interference), and the decay rate corresponds to level 2 (weak interference), it means that the longitudinal energy difference is obvious, but the decay intensity has not reached a medium degree. At this time, level 3 is still taken to ensure that potential risks are not missed.

[0186] Thus, in complex environments such as foggy days / night, the risks of the construction early warning zone can be fully identified, and the intelligent leap of the system from visible to judgeable and from judgeable to warning is finally realized, thereby guaranteeing the safety of traffic and personnel in the construction area.

[0187] The above describes the embodiments of the present embodiment, but the present embodiment is not limited to the specific embodiments described above, and the specific embodiments described above are only illustrative but not limiting, and a person of ordinary skill in the art can make many forms under the inspiration of the present embodiment, which all belong to the protection of the present embodiment.

Claims

1. An online monitoring and early warning system for intelligent transportation engineering construction, characterized in that, include: The data acquisition module acquires single-frame image images of the road construction early warning zone; The image unfolding module converts single-frame images into unfolded images of the road plane based on camera calibration parameters; The image segmentation module marks a long strip of working area set along the center strip of the construction lane on the unfolded image. The long strip of working area covers the near-distance sub-region and far-distance sub-region of the road construction early warning zone. The image transformation module performs a two-dimensional Fourier transform on the long strip working band region to obtain the two-dimensional Fourier amplitude spectrum; The sector extraction module divides the road into a first sector and a second sector based on the road's driving direction; wherein the first sector is aligned with the road's driving direction, and the second sector is orthogonal to the road's driving direction, including: Determine the half-width of the angle, with the unit being radians; calculate the road driving direction angle based on the lateral and longitudinal components of the direction vector of the construction lane. Based on the cosine and sine values ​​of the road driving direction angle, a road driving direction unit vector is constructed. The horizontal component of this road driving direction unit vector is the cosine value of the road driving direction angle, and the vertical component is the sine value of the road driving direction angle. Based on the sine and cosine values ​​of the road driving direction angle, an orthogonal unit vector of the road driving direction is constructed. The horizontal component of this orthogonal unit vector of the road driving direction angle is the negative value of the sine value of the road driving direction angle, and the vertical component is the cosine value of the road driving direction angle. For each frequency point in the two-dimensional Fourier amplitude spectrum, calculate the frequency vector of that frequency point. The horizontal component of the frequency vector is the horizontal frequency count of that frequency point minus half of the total number of horizontal pixels, and the vertical component of the frequency vector is the vertical frequency count of that frequency point minus half of the total number of vertical pixels. For each frequency vector, if the length of the frequency vector is not zero, the frequency vector is normalized to obtain the unit vector of the frequency direction at that frequency point. All frequency points in the two-dimensional Fourier amplitude spectrum that satisfy the first preset condition are formed into the first sector; wherein, the first preset condition is: the dot product of the unit vector in the frequency direction and the unit vector in the road driving direction is not less than the half-width cosine value of the angle. All frequency points in the two-dimensional Fourier amplitude spectrum that satisfy the second preset condition are formed into a second sector; wherein, the second preset condition is: the dot product of the unit vector in the frequency direction and the unit vector orthogonal to the road driving direction is not less than the half-width cosine value of the angle. The energy calculation module calculates the energy values ​​of the first and second sectors within a preset medium spatial frequency range, obtaining the energy in the first direction and the energy in the second direction, including: Set the lower limit and upper limit of the spatial frequency radius; For each frequency point in the two-dimensional Fourier amplitude spectrum, calculate the spatial frequency radius of that frequency point. The spatial frequency radius is the distance from that frequency point to the frequency center. The horizontal coordinate of the frequency center is half of the total number of horizontal pixels, and the vertical coordinate is half of the total number of vertical pixels. The intermediate spatial frequency range is defined as the set of all frequency points in the two-dimensional Fourier amplitude spectrum whose spatial frequency radius is not less than the lower limit of the spatial frequency radius and not greater than the upper limit of the spatial frequency radius. The intersection of the first sector and the intermediate spatial frequency range is calculated to obtain the first directional frequency band; The intersection of the second sector and the intermediate spatial frequency range is calculated to obtain the second directional frequency band; The energy in the first direction is obtained by calculating the sum of squares of the two-dimensional Fourier amplitude spectrum at all frequency points within the first direction frequency band. The energy in the second direction is obtained by calculating the sum of squares of the two-dimensional Fourier amplitude spectrum at all frequency points within the second-direction frequency band. The attenuation calculation module determines the directional energy attenuation rate based on the energy in the first direction and the energy in the second direction, including: setting a stability constant, which is a positive number; Calculate the directional energy ratio, which is the sum of the energy in the first direction and the stability constant, divided by the sum of the energy in the second direction and the stability constant; The ratio of directional energies is taken as the natural logarithm. Taking the negative of the natural logarithm yields the directional energy decay rate; The risk warning module determines the level of directional energy ratio and directional energy attenuation rate of a single frame based on a preset threshold range, and generates the visibility level output of the monitoring screen, including: Obtain the energy ratio and energy decay rate in a single frame direction; Set a ratio threshold of high, medium, and low, and set the ratio threshold of high to be greater than the ratio threshold of medium, which is greater than the ratio threshold of low and greater than zero. Set attenuation rate thresholds to low, medium, and high, with the low attenuation rate threshold being less than or equal to the medium attenuation rate threshold and less than the high attenuation rate threshold. Define four visibility level values, where the first visibility level value is less than the second visibility level value, which is less than the third visibility level value, which is less than the fourth visibility level value. When the directional energy ratio is greater than or equal to the ratio threshold, it corresponds to the first visibility level value; When the directional energy ratio is greater than or equal to the ratio threshold and less than the ratio threshold, it corresponds to the second visibility level value; When the directional energy ratio is greater than or equal to the ratio threshold (low) and less than the ratio threshold (middle), it corresponds to the third visibility level value. When the directional energy ratio is less than the ratio threshold, it corresponds to the fourth visibility level value; When the directional energy attenuation rate is less than the attenuation rate threshold, it corresponds to the first visibility level value; When the directional energy attenuation rate is greater than or equal to the low attenuation rate threshold and less than the medium attenuation rate threshold, it corresponds to the second visibility level value. When the directional energy attenuation rate is greater than or equal to the attenuation rate threshold and less than the attenuation rate threshold, it corresponds to the third visibility level value. When the directional energy attenuation rate is greater than or equal to the attenuation rate threshold, it corresponds to the fourth visibility level value; The larger of the visibility level value obtained by the ratio of energy through the direction and the visibility level value obtained by the energy attenuation rate through the direction is taken as the visibility level output result of the monitoring screen.

2. The online monitoring and early warning system for intelligent transportation engineering construction according to claim 1, characterized in that, Based on camera calibration parameters, a single-frame image is converted into an unfolded image of the road plane, including: Determine the camera intrinsic parameter matrix, camera extrinsic parameter rotation matrix, camera extrinsic parameter translation vector, road surface normal vector, and directed distance from the road surface to the camera coordinate system; Based on the determined camera intrinsic parameter matrix, camera extrinsic parameter rotation matrix, camera extrinsic parameter translation vector, road surface normal vector, and directed distance from the road surface to the camera coordinate system, the homography matrix is ​​calculated. Obtain the homogeneous coordinates of each image pixel in a single image frame, and map the homogeneous coordinates of each image pixel using a homography matrix to obtain the homogeneous coordinates of the corresponding points in the unfolded road plane image. Based on the inverse of the homography matrix, the homography coordinates of each corresponding point in the road plane unfolded image are inversely mapped to obtain the homography coordinates of each corresponding point in a single frame of the image. The homogeneous coordinates of corresponding points in the road planar unfolded image are converted into non-homogeneous coordinates; wherein, the non-homogeneous coordinates are obtained by dividing the first and second components of the corresponding homogeneous coordinates by the third component; The corresponding pixel values ​​in a single image frame are obtained based on the non-homogeneous coordinates of each corresponding point. The road plane unfolded image is then resampled to form the unfolded image of the road plane.

3. The online monitoring and early warning system for intelligent transportation engineering construction according to claim 2, characterized in that, Mark the elongated work zone area along the center strip of the construction lane on the unfolded image. The elongated work zone area covers the near-field and far-field sub-regions of the road construction early warning zone, including: Set the coordinates of the center point of the near sub-region, the coordinates of the center point of the far sub-region, the lane width, the meter-per-pixel scale, the identity matrix, the distance of the lower boundary of the near sub-region, the distance of the upper boundary of the near sub-region, the distance of the lower boundary of the far sub-region, and the distance of the upper boundary of the far sub-region; Based on the coordinates of the center point of the near sub-region and the center point of the far sub-region, the direction vector of the construction lane and the length of the center strip are calculated. The direction vector of the construction lane is used to determine the longitudinal direction of the long strip work area, and the length of the center strip is used to limit the longitudinal range of the long strip work area. For each point in the unfolded image, the longitudinal projection scalar is calculated based on the construction lane direction vector and the difference between the coordinates of the point and the center point of the nearby sub-region. The longitudinal projection scalar is used to measure the longitudinal position of the point on the center zone of the construction lane. Based on the identity matrix, the direction vector of the construction lane, and the difference between the coordinates of the point and the center point of the nearby sub-region, the lateral distance scalar is calculated. The lateral distance scalar is used to measure the lateral distance from the point to the center zone of the construction lane. The elongated working strip region is defined as the area in the unfolded image where the longitudinal projection scalar is between 0 and the length of the central strip, and the lateral distance scalar does not exceed half the width of the lane. Divide the distance between the lower and upper boundaries of the near sub-regions by meters per pixel to obtain the vertical pixel boundaries of the near sub-regions. The portion of the longitudinal projection scalar within the long strip working area between the longitudinal pixel boundaries of the near-distance sub-region is the near-distance sub-region; Divide the distance between the lower and upper boundaries of the distant sub-regions by meters per pixel to obtain the vertical pixel boundaries of the distant sub-regions. The portion of the longitudinal projection scalar within the long strip working area that lies between the longitudinal pixel boundaries of the distant sub-region is the distant sub-region.

4. The online monitoring and early warning system for intelligent transportation engineering construction according to claim 3, characterized in that, Performing a two-dimensional Fourier transform on the elongated working band region yields a two-dimensional Fourier amplitude spectrum, including: Determine the total number of horizontal and vertical pixels in the long strip working area; The elongated working strip region is represented as a grayscale function, which uses the horizontal and vertical pixel positions within the elongated working strip region as variables; Set the horizontal frequency count, vertical frequency count, and imaginary unit; Based on the total number of horizontal pixels, the total number of vertical pixels, the grayscale function, the horizontal frequency count, the vertical frequency count, and the imaginary unit, a two-dimensional discrete Fourier transform is performed on the long strip working area to obtain the complex spectrum of the two-dimensional discrete Fourier transform. Calculate the absolute value of the complex spectrum of the two-dimensional discrete Fourier transform to obtain the two-dimensional Fourier amplitude spectrum.

Images