阅读说明：本技术 一种建构三维图像的道路构造深度检测方法、系统及装置 (Road construction depth detection method, system and device for constructing three-dimensional image ) 是由 窦杨柳 赵广钰 帅进 邹地发 于 2020-12-18 设计创作，主要内容包括：本发明提供一种建构三维图像的道路构造深度检测方法、系统及装置,属于道路检测技术领域。本发明检测方法包括如下步骤：获取建构三维图像的原始数据；对原始数据进行预处理,得到预处理后的高程数据；对高程数据进行校验库回归修正；对各个点修正后的高程值与校验库做标准差分析,得到的残差值作为各个点的最终高程值；依据最终高程值建构出路面构造深度。本发明还提供一种实现所述建构三维图像的道路构造深度检测方法的系统及道路构造深度检测装置。本发明的有益效果为：检测结果精度更高,更加准确。(The invention provides a road structure depth detection method, a system and a device for constructing a three-dimensional image, and belongs to the technical field of road detection. The detection method comprises the following steps: acquiring original data for constructing a three-dimensional image; preprocessing the original data to obtain preprocessed elevation data; performing check library regression correction on the elevation data; performing standard deviation analysis on the corrected elevation value of each point and a check library, and taking the obtained residual error value as the final elevation value of each point; and constructing the pavement structure depth according to the final elevation value. The invention also provides a system for realizing the road construction depth detection method for constructing the three-dimensional image and a road construction depth detection device. The invention has the beneficial effects that: the detection result has higher precision and is more accurate.)

1. A road structure depth detection method for constructing a three-dimensional image is characterized by comprising the following steps:

acquiring original data for constructing a three-dimensional image;

preprocessing the original data to obtain preprocessed elevation data;

step three: performing check library regression correction on the elevation data;

step four: performing standard deviation analysis on the corrected elevation value of each point and a check library, and taking the obtained residual error value as the final elevation value of each point;

step five: and constructing the pavement structure depth according to the final elevation value.

2. The method of claim 1, wherein the method comprises: in the first step, the raw data for constructing the three-dimensional image includes data directly reported by the laser sensor, the image sensor and the distance sensor, or data directly reported by the radar sensor, the image sensor and the distance sensor.

3. The method of claim 2, wherein the method comprises: and in the second step, the data of the laser sensor or the radar sensor in the raw data is subjected to de-jitter and filtering pretreatment.

4. The method as claimed in claim 3, wherein the method comprises: the preprocessing method comprises a Kalman filter, an amplitude limiting filtering method, a median average filtering method, a first-order lag filtering method, a weighted recursive average filtering method, an arithmetic mean filter, a minimum mean square filter, a jitter eliminating filtering method, an amplitude limiting jitter eliminating filtering method or a median filter.

5. The method of claim 1, wherein the method comprises: the processing method in the third step and the fourth step comprises a linear fitting method, a least square method, a gradient descent method or a standard deviation screening method.

6. A system for implementing the method for detecting road structure depth for constructing three-dimensional image according to any one of claims 1 to 5, wherein: the system comprises an acquisition module, a display module and a display module, wherein the acquisition module is used for acquiring original data for constructing a three-dimensional image;

a preprocessing module: the elevation data preprocessing module is used for preprocessing the original data to obtain preprocessed elevation data;

a correction module: the elevation data processing system is used for carrying out check library regression correction on the elevation data;

a standard difference analysis module: the standard deviation analysis is carried out on the corrected elevation values of all the points and the check library, and the obtained residual error value is used as the final elevation value of each point;

constructing a module: and constructing the road surface construction depth according to the final elevation value.

7. A road structure depth detection device which characterized in that: the system comprises a laser sensor, a distance sensor, an image sensor, data acquisition equipment and a data processor, wherein the input end of the data acquisition equipment is connected with the output ends of the laser sensor, the distance sensor and the image sensor, the output end of the data acquisition equipment is connected with the input end of the data processor, and the data processor is provided with the system of claim 6 and is used for constructing the pavement construction depth.

8. The road structure depth detection device according to claim 7, characterized in that: the road structure depth detection device is arranged on a vehicle, and when the vehicle travels a certain distance, the distance sensor respectively sends trigger signals to the laser sensor, the image sensor and the data acquisition equipment; meanwhile, the image sensor, the laser sensor and the distance sensor can simultaneously transmit data to the data acquisition equipment.

9. The road structure depth detection device according to claim 7 or 8, characterized in that: the laser sensor is replaced with a radar sensor.

Technical Field

The present invention relates to a road detection technology, and more particularly, to a road structure depth detection method and system using a three-dimensional image composed of a point laser, a line laser, a linear radar, an area-array radar, or an area-array laser, and a high-speed camera, a line camera, or an area camera, and a road structure depth detection device for implementing the road structure depth detection method for constructing a three-dimensional image.

Background

The current detection methods for measuring the road pavement structure depth are a sand paving method and a vehicle-mounted laser detection method. The sand paving method needs a small amount of equipment for measuring the road structure depth, but the sand paving method needs a large amount of manpower and material resources, is time-consuming and labor-consuming to implement, and is interfered by human factors to reduce the detection precision, so that the sand paving method is used by a small number of people. The vehicle-mounted laser method or the vehicle-mounted radar method is simple and convenient to operate, is not interfered by human factors, light and shadow, oil stains and the like, has the characteristics of convenience in operation and high reliability, and is gradually and widely used in airport runways, high-speed roads and multi-level roads. However, the vehicle-mounted laser detection method or the vehicle-mounted radar detection method has limitations on vehicle speed, and the number of times of scanning areas is different under different vehicle speeds, so that proper compensation is required to be performed through an algorithm according to different vehicle speeds, and the correct measurement value is corrected. In recent years, image processing methods and three-dimensional detection methods are used to perform three-dimensional stereo construction research on road surface features and accurately acquire road construction depth.

The detection modes used by different road construction depths and surface features do not have unified standards at present, and the surface features and the construction depths of different road pavements have inconsistent results due to different detection modes. The measured value has deviation due to different road gradient, surface roughness influence and noise resistance of the sensor, and the data collected by the sensor needs to be further corrected by using a reasonable data processing mode to realize high-precision measured data.

Disclosure of Invention

In order to solve the problems in the prior art, the invention provides a road structure depth detection method and a system for constructing a three-dimensional image by using laser or radar, and further provides a road structure depth detection device for realizing the road structure depth detection method for constructing the three-dimensional image.

The detection method comprises the following steps:

acquiring original data for constructing a three-dimensional image;

preprocessing the original data to obtain preprocessed elevation data;

step three: performing check library regression correction on the elevation data;

step four: performing standard deviation analysis on the corrected elevation value of each point and a check library, and taking the obtained residual error value as the final elevation value of each point;

step five: and constructing the pavement structure depth according to the final elevation value.

The invention is further improved, in the first step, the raw data for constructing the three-dimensional image includes data directly reported by the laser sensor, the image sensor and the distance sensor, or data directly reported by the radar sensor, the image sensor and the distance sensor.

The invention is further improved, in the step two, the data of the laser sensor or the radar sensor in the raw data is subjected to de-jitter and filtering pretreatment.

The invention is further improved, and the preprocessing method comprises a Kalman filter, an amplitude limiting filtering method, a median average filtering method, a first-order lag filtering method, a weighted recursive average filtering method, an arithmetic mean filter, a least mean square filter, a jitter eliminating filtering method, an amplitude limiting jitter eliminating filtering method or a median filter.

The invention is further improved, and the processing method of the third step and the fourth step comprises a linear fitting method, a least square method, a gradient descent method or a standard deviation screening method.

The invention also provides a system for realizing the detection method, which comprises an acquisition module, a detection module and a detection module, wherein the acquisition module is used for acquiring the original data for constructing the three-dimensional image;

a preprocessing module: the elevation data preprocessing module is used for preprocessing the original data to obtain preprocessed elevation data;

a correction module: the elevation data processing system is used for carrying out check library regression correction on the elevation data;

a standard difference analysis module: the standard deviation analysis is carried out on the corrected elevation values of all the points and the check library, and the obtained residual error value is used as the final elevation value of each point;

constructing a module: and constructing the road surface construction depth according to the final elevation value.

The invention also provides a road structure depth detection device, which comprises a laser sensor, a distance sensor, an image sensor, data acquisition equipment and a data processor, wherein the input end of the data acquisition equipment is connected with the output ends of the laser sensor, the distance sensor and the image sensor, the output end of the data acquisition equipment is connected with the input end of the data processor, and the data processor is provided with the system for constructing the road structure depth.

The invention is further improved, the road structure depth detection device is arranged on a vehicle, and when the vehicle travels a certain distance, the distance sensor respectively sends out trigger signals to the laser sensor, the image sensor and the data acquisition equipment; meanwhile, the image sensor, the laser sensor and the distance sensor can simultaneously transmit data to the data acquisition equipment.

The invention is further improved, and the laser sensor is replaced by a radar sensor.

Compared with the prior art, the invention has the beneficial effects that: the elevation value with high stability, high precision and high information degree can be obtained, so that the detection result is higher in precision and more accurate.

Drawings

FIG. 1 is a flow chart of the method of the present invention;

fig. 2 is a block diagram of the road structure depth detection device according to the present invention.

Detailed Description

The present invention will be described in further detail with reference to the accompanying drawings and examples.

As shown in fig. 1, the detection method of the present invention for constructing a road structure depth of a three-dimensional image based on laser or radar includes the following steps:

step one, acquiring original data for constructing a three-dimensional image. The original data are directly acquired by the sensors and are data without processing, so that the deviation of the measured values caused by different anti-noise capabilities of the sensors can be effectively avoided, and the consistency is better.

And step two, performing de-jitter and filtering pretreatment on the original data to obtain the pretreated elevation data. In this embodiment, besides the hardware plus anti-jitter method, the method of pre-processing the data acquired by the laser sensor or the radar sensor by de-jitter filtering and the method of using software as the de-jitter filtering function are not limited to using Kalman Filter (Kalman Filter), Kalman-Bucy Filter, hybrid Kalman Filter, Extended Kalman Filter (Extended Kalman Filter), frequency weighted Kalman Filter, amplitude-limiting filtering method, median-value average filtering method, first-order lag filtering method, weighted recursive average filtering method, arithmetic mean Filter, least mean square Filter, jitter-eliminating filtering method, amplitude-limiting jitter-eliminating filtering method, median Filter, etc. in software. No matter what kind of software filter is used as the method of de-jitter filtering function, the original purpose of the invention is not violated, and the most effective way of de-jitter is that after some series of initial values pass through the discrete dynamic system, parameters in the discrete system are continuously and iteratively corrected, so that the predicted value tends to approach the next measurement value.

The present example describes the processing method of the present invention in detail using a kalman filter. Using Kalman filtering to remove jitter and carrying out filtering pretreatment has 4 steps, including predicting the next state of the system by using a system process model, updating covariance matrix parameters, combining the predicted values and the optimized values of the measurement values, and correcting the difference matrix parameters, and finally substituting the finally corrected covariance matrix into the system process model again, so that the measurement and the prediction can be continuously corrected, and the predicted values tend to approach the next measurement values.

Step S2.1: the lattice data measured for each laser is given to a linear random differential equation (1.1)

X[k]＝AX[k-1]+BU[k]+W[k] (1.1)

Plus the system measurements:

Z[k]＝HX[k]+V[k] (1.2)

wherein X [ k ] is the system state at time k, and U [ k ] is the control quantity of the system at time k. A and B are system parameters, which are both matrices for a multi-model system. Z [ k ] is the measured value at time k, H is a parameter of the measurement system, and H is a matrix for a multi-measurement system. Wk and Vk represent the noise of the process and measurement, respectively. This example is assumed to be white gaussian noise (whitegaussian noise) and their covariance (covariance) is Q, R, respectively.

P represents the covariance (covariance), which is the covariance corresponding to X [ k | k-1], and the covariance is calculated as:

P[k|k-1]＝AP[k-1|k-1]A′+Q (2)

in the formula (2), P [ k | k-1] is a covariance corresponding to X [ k | k-1], A' represents a transposed matrix of A, Q is a covariance of a system process, and P [ k-1| k-1] is a covariance corresponding to X [ k-1| k-1] which is data measured from a current state (at k-1) to predict data measured from a next state (at k). This step is a prediction of the current K-time state. This example refers to the data measured at the current K.

Step S2.2: equation (2) represents the prediction result of the current state, and then the present example obtains the optimized estimated value X [ k | k ] of the current state (k) by collecting the measured values of the current state and combining the predicted values and the measured values:

X[k|k]＝X[k|k-1]+Kg[k](Z[k]-HZ[k|k-1]) (3)

wherein Kg is Kalman gain, and the calculation formula of Kg is as follows:

where z (k) is the measurement value at time k, H is a parameter of the measurement system, H is a matrix for a multi-measurement system, H' refers to the transposed matrix of H, w (k) and v (k) represent the noise of the process and measurement, respectively. They are assumed to be White Gaussian Noise (White Gaussian Noise) and their covariance (covariance) is Q, R, respectively (here we assume that they do not change with changes in system state). The present invention has thus far obtained an optimal estimate x k for the k state.

Step S2.3: in order to make the kalman filter continuously operate until the system process is finished, the covariance of x [ k | k ] in the k state is updated in this example, and the calculation formula is:

P[k|k]＝(I-Kg[k]H)P[k|k-1] (5)

where I is the matrix of [1], I ═ 1 for single model single measurements. When the system enters the k +1 state, P [ k | k ] is P [ k-1| k-1] of equation (2).

Step S2.4: the measured values are again cycled through steps S1.1 and S1.2, and the algorithm can proceed with autoregressive operation, so as to obtain the preprocessed elevation data at each time point.

And step three, performing check library regression correction on the elevation data.

Step four: and performing standard deviation analysis on the corrected elevation value of each point and the check library to obtain a residual error value serving as a final elevation value of each point.

In the third step and the fourth step of the present embodiment, algorithms such as a linear fitting method, a least square method, a gradient descent method, a standard deviation screening method, and the like can be uniformly used. In this example, steps S3 and S4 use a linear fit method plus a least squares method to calculate the residual error of the laser measured elevation data after de-jitter filtering according to step S1.

Step S3: and performing data regression correction on the data subjected to de-jitter and filtering pretreatment.

Step S3.1: de-jittered and filtered pre-processed elevation dataWhereinIs oneThe matrix is a matrix of a plurality of matrices,is a value representing the elevation after the k-th filtering processThe complete formula is as follows:

wherein the content of the first and second substances,the elevation value after the kth filtering processing is represented, T represents the number of elevation value points, m represents the number of elevation point points in a filtering window, i represents the ith elevation point, and j represents a variable to replace the expression of i + m-1.

Step S3.2: the elevation data matrix in step S3.1 needs to be mappedAnd performing linear fitting with an optimal data fitting line of the elevation point of the check library, wherein the optimal data fitting line formula of the elevation point is set as follows:

where i denotes the ith elevation point, i ═ 1 … … n, and z denotes the regression value of the ith elevation point.Andcoefficients of a best data fit linear equation for the elevation data measured over time for the ith elevation point average, respectively.

Linear fitting equation (9):

wherein the content of the first and second substances,represents the average of the n elevation points,represents the ith filtered elevation value,representing the mean of the n filtered elevation values, i.e.

Step S4: residual error analysis is carried out on the elevation values measured at all the points and the check library, and the obtained residual error values are used as the values of all the points;

the average value of the elevation values obtained by the formula (13)Then according to the obtained elevation value of each i point of the fitting regression lineThen according to the formula:

calculating the residual values to obtain the final elevation value, wherein,representing the residual of the ith elevation value.

Step S5: and constructing the road surface structure depth by using the residual error of the elevation values of all the points.

The road surface structure depth can be constructed by calculating the height residuals according to the formula (14) in step S4.

As shown in fig. 2, in order to obtain a high-stability, high-precision and high-information-degree elevation value, this example further provides a road structure depth detection device, which includes a laser sensor, a distance sensor, an image sensor, a data acquisition device and a data processor, wherein an input end of the data acquisition device is connected to output ends of the laser sensor, the distance sensor and the image sensor, an output end of the data acquisition device is connected to an input end of the data processor, and the data processor is provided with the system for constructing a road structure depth.

The distance sensor of this example is not limited to an axle encoder, an inertial navigation system, a GPS system, and any device capable of detecting a distance or positioning. The laser sensor is not limited to a point laser, a line laser, and a 3-dimensional laser. The image sensor is not limited to a three-dimensional camera, an area-array camera, a line camera, and any sensor or device capable of acquiring an image. The sensors referred to above are regarded as prior art and, when applied to the present device, should be protected by the method of the present invention for constructing the depth of the road pavement structure.

The road structure depth detection device of the present example operates on the following principle:

when the vehicle travels a certain distance, the distance sensor respectively sends out a trigger signal to the laser sensor (or the radar sensor), the image sensor and the data acquisition equipment; meanwhile, the image sensor, the laser sensor and the distance sensor (or the radar sensor) can simultaneously transmit data to the data acquisition equipment.

2. The data acquisition equipment collects data transmitted by the image sensor, the laser sensor (or the radar sensor) and the distance sensor, and stores the data correspondingly according to actual use requirements. For example: the data can be transmitted to the data processor system according to different requirements of different distances, such as 0.2 meter, 0.5 meter, 1 meter, 2 meters, 5 meters and the like, and the data is not processed at the moment.

3. The data acquisition device transmits the acquired data to the data processor system and allows the data processor system to perform further processing.

When the vehicle runs into the Poiselle section, in order to prevent the inaccurate elevation value from being obtained, the hardware anti-jitter and software de-jitter filtering modes are firstly performed, and the accurate elevation value is obtained and then processed by a subsequent algorithm. However, the use of software for de-jitter filtering may have different accuracy due to different software log processing methods. The system architecture described in the invention corrects the obtained numerical value, a software de-jitter filtering mode and a regression value matched with a check library, and the obtained elevation value is analyzed with a standard deviation of the check library to obtain a residual value, wherein the residual value can be used as an accurate elevation value at each point, and the road pavement depth can be constructed by the accurate elevation value. The invention adopts the data after the jitter preprocessing to obtain the elevation value with higher precision than the existing method which only uses the check library. The road pavement structure depth with the elevation value can be constructed, the structure depth of the required distance can be set in a manual mode according to the distance sensor, and the method is more flexible in practical application.

The above-described embodiments are intended to be illustrative, and not restrictive, of the invention, and all changes that come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.