CN104964708A - Pavement pit detecting method based on vehicular binocular vision - Google Patents

Abstract

The invention is a road surface pothole detection method based on vehicle-mounted binocular vision. The detection method uses the vibration sampling information of the vibration sensor to take a reasonable threshold to screen out the road surface potholes, and combines GPS information to provide the location information of the road surface potholes. When the wheel leaks pressure on the road pothole, the GPS information at that time can be recorded through the button-type manual external trigger device. The binocular camera system samples the test road surface and outputs it to the computer. After obtaining the image of the road pothole, the image is preprocessed and image The processing algorithm obtains the area and depth information of road potholes. By adopting the technical solution of the invention, not only can the positioning of road potholes be completed, but also the traditional image processing algorithm is optimized in terms of algorithms, and the functions of calculating the area and depth of road potholes can be accurately realized.

Description

A road surface pothole detection method based on vehicle-mounted binocular vision

technical field

The invention relates to the field of road detection and image processing, in particular to a detection method for measuring the position, area and depth information of road potholes based on binocular vision and using an image processing algorithm.

Background technique

In recent years, economic development has led to an increase in the number of vehicles. With the accumulated wear and tear of the road surface, the appearance of potholes is inevitable. Therefore, the importance of road pothole detection for vehicle driving is self-evident. Pothole inspection is an important part of pavement quality inspection. The development of digital image processing technology makes traditional pavement inspection work more intelligent and mechanized.

However, pavement potholes are a special form of pavement damage. Traditional road surface measurement methods, such as bump accumulators and laser rangefinders, can only obtain the curve of the longitudinal section of the road surface, and cannot locate the position of the pavement potholes, let alone obtain its spatial information. The road surface image processing based on the traditional edge detection algorithm cannot obtain the edge of the irregular closed area like the road pothole. The traditional monocular camera sampling the road surface image cannot measure the road surface due to the limitation of its two-dimensional plane dimension. Spatial 3D information of potholes.

Contents of the invention

In order to solve the defect that the traditional method cannot measure the space information of road potholes, the present invention combines GPS, binocular cameras, vibration sensors and other hardware devices, and uses the vehicle-mounted method to locate the position of the potholes, collect images of the potholes, and use The image algorithm calculates the area and depth information of the pit.

In order to achieve the above-mentioned technical purpose and achieve the above-mentioned technical effect, the present invention is realized through the following technical solutions:

A road surface pothole detection method based on vehicle-mounted binocular vision. The hardware based on the method includes a main control chip, and the main control chip is respectively connected with a binocular camera system, a GPS positioning system, a vibration sensor, an artificial external trigger device and a computer. , wherein the binocular camera system is installed on the roof of the vehicle to collect road images, the GPS positioning system and the binocular camera system are fixed on the roof of the vehicle together, and the vibration sensor is fixed on the dashboard of the vehicle, The manual external trigger device is fixed on the co-pilot panel, and the computer calculates the area of road surface pits and the depth of road surface pits according to the sampling data returned by the main control chip:

The method for measuring and calculating the area of potholes on the road surface comprises the following steps:

Step 1.1) Receive the road sampling image and calibrate the area occupied by each pixel at the same time;

Step 1.2) Denoise the image with an adaptive median filter;

Step 1.3) Adopt fast edge detection based on region growing rules;

Step 1.4) Check whether it is a closed edge and make a judgment. If not, it means that it is not a road pothole, record the GPS information, and jump to step 1.1), if yes, go to the following step 1.5);

Step 1.5) Count the total number of pixels on the boundary and the total number of pixels within the boundary;

Step 1.6) Combine the calibrated area of each pixel to finally obtain the area of the road pothole;

The method for calculating the depth of the pavement pothole comprises the following steps:

Step 2.1) Receive the road sampling image;

Step 2.2) Denoise the image with an adaptive median filter;

Step 2.3) Adopt fast edge detection based on region growing rules;

Step 2.4) Detect whether it is a closed edge and make a judgment, if not, it means that it is not a road surface pothole, go to the following step 2.5), if it is, then jump to step 2.6);

Step 2.5) record GPS information, and jump to step 2.1);

Step 2.6) Obtain the closed edge of the binocular image and obtain the binocular image at the same moment;

Step 2.7) Check whether the two images have the same image point and make a judgment, if not, then jump to step 2.5), if so, then go to the following step 2.8);

Step 2.8) Carry out image point matching with the same name;

Step 2.9) Calculate the road tangent plane;

Step 2.10) Calculate the distance between each point in the edge and the tangent plane of the road surface;

Step 2.11) Take the peak value as the depth of the road surface pothole.

Further, the binocular camera system is installed on the roof of the vehicle at a fixed angle.

Further, the vibration sensor adopts a three-axis accelerometer.

Further, in the step 1.4) and step 2.5), the GPS information at that time is recorded by a manual external trigger device when the wheel leaks pressure on the road surface pothole.

Further, the manual external trigger device is a button type.

The beneficial effects of the present invention are:

The present invention can complete the positioning of road potholes through the construction of hardware, optimizes the traditional image processing algorithm in terms of algorithms, and can accurately realize the calculation function of the area and depth of road potholes. These data can be used for road quality supervision to a large extent. The department provides a new detection technology, which can also provide support for future road safety protection, which has important engineering practical significance.

Description of drawings

Fig. 1 is a schematic diagram of the system structure of the present invention;

Figure 2 is a flow chart for calculating the area of road pits;

Fig. 3 is the flow chart of measuring and calculating the depth of the pavement pothole;

Fig. 4 is a three-dimensional coordinate diagram of different image point spaces taken in the embodiment of the present invention.

Detailed ways

The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments.

Referring to Fig. 1, a road surface pothole detection method based on vehicle-mounted binocular vision, the hardware based on the method includes a main control chip, and the main control chip is respectively connected with a binocular camera system, a GPS positioning system, a vibration sensor, The artificial external trigger device and computer are characterized in that the binocular camera system is installed on the roof of the vehicle to collect road images, the GPS positioning system is fixed on the roof of the vehicle together with the binocular camera system, and the vibration sensor is fixed on the roof of the vehicle. In the instrument panel of the vehicle, the manual external trigger device is fixed on the co-pilot panel, and the computer calculates the area of the road surface pothole and the depth of the road surface pothole according to the sampling data returned by the main control chip:

With reference to shown in Figure 2, described pavement pothole area calculation method comprises the following steps:

Step 1.1) Receive the road sampling image and calibrate the area occupied by each pixel at the same time;

Step 1.2) Denoise the image with an adaptive median filter;

Step 1.3) Adopt fast edge detection based on region growing rules;

Step 1.4) Check whether it is a closed edge and make a judgment. If not, it means that it is not a road pothole, record the GPS information, and jump to step 1.1), if yes, go to the following step 1.5);

Step 1.5) Count the total number of pixels on the boundary and the total number of pixels within the boundary;

Step 1.6) Combine the area occupied by each calibrated pixel to finally obtain the area of the road pothole. Specifically, combine the calibrated parameter information, according to the statistical pothole edge pixel coordinates, the total number of pixels within the technical pothole range, and then obtain the road pothole actual area;

With reference to shown in Figure 3, described pavement pothole depth measuring method comprises the following steps:

Step 2.1) Receive the road sampling image;

Step 2.2) Denoise the image with an adaptive median filter;

Step 2.3) Adopt fast edge detection based on region growing rules;

Step 2.4) Detect whether it is a closed edge and make a judgment, if not, it means that it is not a road surface pothole, go to the following step 2.5), if it is, then jump to step 2.6);

Step 2.5) record GPS information, and jump to step 2.1);

Step 2.6) Obtain the closed edge of the binocular image and obtain the binocular image at the same moment;

Step 2.7) Check whether the two images have the same image point and make a judgment, if not, then jump to step 2.5), if so, then go to the following step 2.8);

Step 2.8) Carry out image point matching with the same name;

Step 2.9) Calculate the road tangent plane;

Step 2.10) Calculate the distance between each point in the edge and the tangent plane of the road surface;

Step 2.11) Take the peak value as the depth of the road surface pothole.

The binocular camera system is installed on the roof of the vehicle at a fixed angle.

The vibration sensor uses a three-axis accelerometer.

In the above step 1.4) and step 2.5), when the wheel leaks pressure on the road surface pothole, the GPS information at that time is recorded through the manual external trigger device.

The manual external trigger device is a button type.

In the present embodiment, table 1 represents in the test process, obtains the area of four pavement potholes, and utilizes the method of the present invention to also be able to accurately measure the object area of some irregular shapes:

Table 1 Measurement data of pit image area (area unit: )

In the present embodiment, table 2 is the three-dimensional coordinate value of different image points:

Table 2 The calculated value of the spatial three-dimensional coordinates based on the maximum performance measure of the correlation coefficient for the matching of the image point with the same name

For different stereo coordinates, it can be represented in three-dimensional space. In Fig. 4, the coordinate system in Fig. 4 is set. Combining with Fig. 4, it can be seen that the vertical coordinate of the point in the middle of Fig. The section is relatively high.

It can be seen from Fig. 4 that the range of the ordinate is 17.122mm-48.772mm. In this way, it is deduced that the depth of the pothole on the road surface that needs to be detected is about 31.65mm.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and changes. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims (5)

1. hole, the road surface based on vehicle-mounted binocular vision groove detection method, the method based on hardware comprise main control chip, described main control chip is connected to binocular camera system, GPS positioning system, vibration transducer, artificial external trigger device and computing machine, it is characterized in that, described binocular camera system is installed on road pavement image on vehicle ceiling and gathers, described GPS positioning system is fixed on vehicle ceiling together with binocular camera system, described vibration transducer is fixed in meter panel of motor vehicle, described artificial external trigger device is fixed on copilot panel, described computing machine carries out road surface hole groove area and the groove depth information measuring and calculating of hole, road surface according to the sampled data that main control chip is passed back:

Hole, described road surface groove area measuring method comprises the following steps:

Step 1.1) receive road surface sampled images, demarcate each pixel area occupied simultaneously;

Step 1.2) to image adaptive median filter noise reduction;

Step 1.3) adopt based on the rapid edge-detection of Rule of Region-growth;

Step 1.4) detect whether be closure edge and differentiate, if not, then illustrate be not road surface hole groove, record GPS information, jump to step 1.1 simultaneously), if so, then go to step 1.5 below);

Step 1.5) sum of all pixels on statistical boundary and the sum of all pixels in border;

Step 1.6) combine the area that each pixel area occupied of demarcating finally obtains hole, road surface groove;

Hole, described road surface groove depth measuring method comprises the following steps:

Step 2.1) receive road surface sampled images;

Step 2.2) to image adaptive median filter noise reduction;

Step 2.3) adopt based on the rapid edge-detection of Rule of Region-growth;

Step 2.4) detect whether be closure edge and differentiate, if not, then illustrate be not road surface hole groove, go to step 2.5 below), if so, then jump to step 2.6);

Step 2.5) record GPS information, jump to step 2.1) simultaneously;

Step 2.6) obtain the closure edge of binocular image and obtain binocular image in the same time;

Step 2.7) check whether two width images have corresponding image points and differentiate, if not, then jump to step 2.5), if so, then go to step 2.8 below);

Step 2.8) carry out corresponding image points coupling;

Step 2.9) calculate section, road surface;

Step 2.10) each point in edge is calculated the distance between section, road surface respectively;

Step 2.11) get the degree of depth of peak value as hole, road surface groove.

2. hole, the road surface based on vehicle-mounted binocular vision according to claim 1 groove detection method, it is characterized in that, described binocular camera system is arranged on vehicle ceiling with fixing angle.

3. hole, the road surface based on vehicle-mounted binocular vision according to claim 1 groove detection method, it is characterized in that, described vibration transducer adopts three axis accelerometer.

4. hole, the road surface based on vehicle-mounted binocular vision according to claim 1 groove detection method, is characterized in that, in described step 1.4) and step 2.5) in, record GPS information at that time when wheel leaks when groove is cheated on pressure road surface by artificial external trigger device.

5. hole, the road surface based on vehicle-mounted binocular vision according to claim 4 groove detection method, it is characterized in that, described artificial external trigger device is button.