CN105224908A - A kind of roadmarking acquisition method based on orthogonal projection and device - Google Patents

Abstract

The invention provides a method and device for collecting road markings based on orthographic projection, and relates to the technical field of images and videos. Wherein, the above-mentioned method includes: using the camera device installed on the acquisition vehicle to collect road images to obtain the target image; performing image processing on the target image to obtain the position of the road edge in the target image; according to the position of the road edge in the target image to determine The position of the road, and based on the position of the road, extract the feature points of the target image to obtain the calibration feature points; obtain the conversion relationship between the camera device coordinate system and the space coordinate system according to the calibration feature points, and generate an inverse perspective transformation matrix; according to the inverse perspective The transformation matrix transforms the image of the target image to obtain the front view of the target image. This method realizes the real-time collection of road markings, and at the same time realizes the automation of calibration, correction, and front view generation, which greatly saves the time for subsequent processing and saves a lot of manpower and material resources.

Description

A method and device for collecting road markings based on orthographic projection

technical field

The invention relates to the field of image and video technology, in particular to a method and device for collecting road markings based on orthographic projection.

Background technique

The collection of ground markings is one of the map information that map information collectors pay more attention to. In the existing collection schemes, in addition to manual collection methods, the method of map collection vehicles is generally used. Traditional collection vehicles need to be professionally installed with various sensors. To collect ground marking lines, a professional camera must be fixed on the vehicle, and there are strict requirements on the angle of the camera. At the same time, manual conversion between the camera coordinate system and the vehicle coordinate system is required. Calibrate. After calibration, if the camera is reinstalled, fixed or even moved, it needs to be recalibrated again. Therefore, in order to ensure the accuracy of the collected data, it is often necessary to recalibrate before collecting. When the transformation relationship between the coordinate system between the camera and the vehicle is clear, the ground image captured by the camera at a certain angle can be converted into an image observed vertically from a top-down perspective (front view).

However, this traditional collection method has many inconveniences. For example, each collection vehicle needs to be specially modified, and cameras are added, which requires a large amount of work. When deployed in large quantities, the time cost and capital cost are relatively large; manual calibration is required, and this method is cumbersome. It is time-consuming, and it cannot be corrected in time when the camera angle changes unexpectedly, and the measurement error is large.

Contents of the invention

The purpose of the present invention is to provide a method and device for collecting road markings based on orthographic projection, which realizes the automation of calibration, correction, and front view generation, greatly saves the time for subsequent processing, and saves a lot of manpower and material resources.

In order to achieve the above purpose, an embodiment of the present invention provides a method for collecting road markings based on orthographic projection, including:

Use the camera device installed on the acquisition vehicle to collect road images and obtain target images;

performing image processing on the target image to obtain the position of the road edge in the target image;

Determining the position of the road according to the position of the road edge in the target image, and extracting feature points from the target image based on the position of the road to obtain marked feature points;

Acquiring the conversion relationship between the coordinate system of the camera device and the space coordinate system according to the calibration feature points, and generating an inverse perspective transformation matrix;

performing picture conversion on the target image according to the inverse perspective transformation matrix to obtain a front view of the target image.

Wherein, after obtaining the front view of the target image, it also includes:

The front view of the target image is displayed and stored in real time.

Wherein, performing image processing on the target image, and obtaining the position of the road edge in the target image comprises:

converting the target image into a grayscale image;

Converting the grayscale image into an edge map showing only the edge of the object;

Obtaining the first straight line feature in the edge graph, and filtering the first straight line feature according to the straight line feature of the road edge of the preset road model, and obtaining the second straight line feature;

If the current target image is the first frame image collected, the second straight line feature is the target straight line feature; otherwise, according to the second straight line feature of the multi-frame images before the current target image, the predicted road edge information is obtained, And combining the predicted road edge information and the second straight line feature of the current target image to determine the target straight line feature;

According to the target straight line feature, the position of the road edge in the target image is acquired.

Further, the step of converting the grayscale image into an edge map showing only the edge of the object includes:

A canny edge detection operator is used to convert the grayscale image into an edge map that only shows the edge of the object.

Further, the step of obtaining the first straight line feature in the edge graph includes:

A random hough transform method is applied to obtain the first straight line feature in the edge graph.

Wherein, the conversion relationship between the coordinate system of the camera device and the space coordinate system is obtained according to the calibration feature points, and the step of generating an inverse perspective transformation matrix includes:

Obtain the actual position of the calibration feature point;

According to the position of the calibration feature point on the target image and the actual position, using an inverse perspective transformation method to determine the conversion relationship between the coordinate system of the camera device and the spatial coordinate system;

An inverse perspective transformation matrix is generated according to the conversion relationship between the camera coordinate system and the space coordinate system.

Wherein, performing image conversion on the target image according to the inverse perspective transformation matrix to obtain a front view of the target image includes:

Crop the target image, and only keep the part below the horizon on the target image;

performing picture conversion on the cropped target image according to the inverse perspective transformation matrix to obtain a fan-shaped picture;

Cutting the fan-shaped picture based on the position of the road edge on the target image to obtain a rectangular picture, the rectangular picture being a front view of the target image.

The embodiment of the present invention also provides a road marking collection device based on orthographic projection, including:

The collection module is used to collect road images using the camera installed on the collection vehicle to obtain target images;

An acquisition module, configured to perform image processing on the target image, and acquire the position of the road edge in the target image;

An extraction module, configured to determine the position of the road according to the position of the road edge in the target image, and perform feature point extraction on the target image based on the position of the road to obtain calibration feature points;

A conversion module, configured to obtain the conversion relationship between the camera coordinate system and the space coordinate system according to the calibration feature points, and generate an inverse perspective transformation matrix;

A generating module, configured to perform picture conversion on the target image according to the inverse perspective transformation matrix, to obtain a front view of the target image.

Further, the road marking acquisition device also includes:

The display storage module is used to display and store the front view of the target image in real time.

Wherein, the acquisition module includes:

The first acquisition submodule is used to convert the target image into a grayscale image;

The second acquisition sub-module is used to convert the grayscale image into an edge image showing only the edge of the object;

The third obtaining sub-module is used to obtain the first straight line feature in the edge map, and filter the first straight line feature according to the straight line feature of the road edge of the preset road model, and obtain the second straight line feature;

The fourth acquisition sub-module is used for if the current target image is the first frame image collected, the second straight line feature is the target straight line feature; otherwise, according to the second straight line of multiple frames of images before the current target image feature, obtaining predicted road edge information, and combining the predicted road edge information and the second straight line feature of the current target image to determine the target straight line feature;

The fifth acquisition sub-module is configured to acquire the position of the road edge in the target image according to the target straight line feature.

Further, the second acquisition submodule includes:

The first conversion module is used to convert the grayscale image into an edge image showing only object edges by using a canny edge detection operator.

Further, the third acquisition submodule includes:

The second conversion module is used to obtain the first straight line feature in the edge graph by applying the random hough transform method.

Wherein, the conversion module includes:

The first conversion submodule is used to obtain the actual position of the calibration feature point;

The second conversion sub-module is used to determine the conversion relationship between the coordinate system of the camera device and the spatial coordinate system by using an inverse perspective transformation method according to the position of the calibration feature point on the target image and the actual position;

The third transformation sub-module is used to generate an inverse perspective transformation matrix according to the transformation relationship between the vehicle coordinate system and the space coordinate system.

Wherein, the generating module includes:

The first generating submodule is used to crop the target image, and only keep the part below the horizon on the target image;

The second generation sub-module is used to perform image conversion on the cropped target image according to the inverse perspective transformation matrix to obtain a fan-shaped image;

The third generation sub-module is configured to cut the fan-shaped picture based on the position of the road edge on the target image to obtain a rectangular picture, and the rectangular picture is a front view of the target image.

The technical solution of the present invention has at least the following beneficial effects:

In the method for collecting road markings based on orthographic projection in the embodiment of the present invention, the road edge is identified by processing the collected target image, and self-calibration is performed through the detection of the marked feature points of the road and the calculation of cumulative sampling correction, Obtain the inverse perspective transformation matrix, and then convert the target image into a front view, realize the real-time collection of road markings, and realize the automation of calibration, correction, and front view generation, which greatly saves the time for subsequent processing and saves a lot Manpower and material resources.

Description of drawings

Fig. 1 shows the flow chart of the basic steps of the road marking collection method based on orthographic projection in an embodiment of the present invention;

Fig. 2 represents the specific flowchart of determining the road edge position in the method for collecting road markings based on orthographic projection in an embodiment of the present invention;

Fig. 3 represents the specific flowchart of generating the inverse transformation matrix in the method for collecting road markings based on orthographic projection in an embodiment of the present invention;

Fig. 4 shows the coordinate mapping relation of generating the inverse transformation matrix in the method for collecting road markings based on orthographic projection in an embodiment of the present invention;

5 shows a specific flow chart of generating a front view of a target image in a method for collecting road markings based on orthographic projection according to an embodiment of the present invention;

FIG. 6 shows a schematic diagram of the composition and structure of a device for collecting road markings based on orthographic projection according to an embodiment of the present invention.

detailed description

In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, the following will describe in detail with reference to the drawings and specific embodiments.

The present invention aims at the problem that vehicles for collecting road markings in the prior art need to be specially refitted, the engineering volume is large, manual calibration is required, which is cumbersome and time-consuming, and it is impossible to make corrections in time when the camera angle changes unexpectedly, and provides a method based on orthographic projection The road marking collection method and device of the present invention process the collected target image to identify the road edge, and perform self-calibration through the calculation of road calibration feature point detection and cumulative sampling correction to obtain an inverse perspective transformation matrix, and then The target image is converted into a front view, which realizes the real-time collection of road markings, and at the same time realizes the automation of calibration, correction, and front view generation, which greatly saves the time for subsequent processing and saves a lot of manpower and material resources.

As shown in Figure 1, an embodiment of the present invention provides a method for collecting road markings based on orthographic projection, including:

Step 1, using the camera installed on the acquisition vehicle to collect road images to obtain target images;

Step 2, performing image processing on the target image to obtain the position of the road edge in the target image;

Step 3, determining the position of the road according to the position of the edge of the road in the target image, and extracting feature points from the target image based on the position of the road to obtain calibration feature points;

Step 4, obtaining the conversion relationship between the coordinate system of the camera device and the space coordinate system according to the calibration feature points, and generating an inverse perspective transformation matrix;

Step 5: Perform picture conversion on the target image according to the inverse perspective transformation matrix to obtain a front view of the target image.

In the above-mentioned embodiments of the present invention, the target image collected by the camera device is processed to obtain the position of the road edge in the target image; wherein, the road includes information such as the road surface and the road edge, and the road edge refers to the road edge; in layman's terms , the road is a surface, and the road edge is two lines; that is, the road (or lane) refers to the passable area between two lane lines (such as between the solid line and the dashed line, between the double dashed lines, etc.); and the road The edge can be roughly understood as a lane line, but because the dotted line of the lane line is intermittent, and the road edge is continuous, the road edge can be understood as two straight lines that limit the passable area of the road.

Further, the position of the road edge in the target image can reflect the angle information between the camera device and the ground; according to the road edge information determined in step 2, the position of the road in the target image can be determined, and several mark feature points; perform step 4, according to the information of the mark feature points, the conversion relationship between the camera coordinate system and the space coordinate system can be obtained, and an inverse perspective transformation matrix is generated; then the target image is converted in combination with the inverse perspective transformation matrix, Get the front view of the target image.

Wherein, the orthographic projection refers to converting the picture captured by the camera device with a certain inclination angle to the ground into a picture directly vertically downward (orthographic projection). The direct output of the method provided by the present invention is the real-time orthophoto picture (front view), which also greatly simplifies the steps of collecting ground markings, realizes the automation of calibration, correction and generation of the front view, and greatly It saves the time for subsequent processing, improves work efficiency, and saves a lot of manpower and material resources.

In the above-mentioned embodiment of the present invention, after step 5, it also includes:

Step 6, displaying and storing the front view of the target image in real time.

Specifically, in the embodiment of the present invention, after the front view is generated, the front view is displayed and stored in real time, so as to achieve the purpose of collecting ground markings and provide basic data for the collection of map information. Preferably, the front view can be displayed on the screen, but it is not limited to this method. It can also be used through the head-up display system, such as mapping the front view to the front windshield, etc., not limited to a fixed method, and it is not used here. A repeat.

It should be noted that the storage of the front view of the target image can be stored in a preset memory or database, so that its data can be recalled later.

In the above-mentioned embodiment of the present invention, as shown in Figure 2, step 2 includes:

Step 21, converting the target image into a grayscale image;

Step 22, converting the grayscale image into an edge map that only shows the edge of the object;

Step 23, obtaining the first straight line feature in the edge map, and filtering the first straight line feature according to the straight line feature of the road edge of the preset road model to obtain the second straight line feature;

Step 24, if the current target image is the first frame image collected, the second straight line feature is the target straight line feature; otherwise, obtain the predicted road according to the second straight line features of multiple frames of images before the current target image Edge information, and combining the predicted road edge information and the second straight line feature of the current target image to determine the target straight line feature;

Step 25: Acquire the position of the road edge in the target image according to the feature of the target line.

In a specific embodiment of the present invention, the method of converting the target image into a grayscale image in step 21 can be changed by changing the attributes of the image, or by some image editing software, such as photoshop, etc., which are not listed here.

Preferably, step 22 includes:

Step 221 , using a canny edge detection operator to convert the grayscale image into an edge map that only shows the edge of the object.

The Canny edge detection operator is a multi-level edge detection algorithm developed by JohnF.Canny in 1986; and Canny created the Computational theory of edge detection to explain the working principle of the Canny edge detection operator. The basic steps of the Canny edge detection operator include a. Denoising; b. Finding the brightness gradient in the image; c. Tracking the edge in the image. The goal of the Canny edge detection operator is to find an optimal edge detection algorithm. The meaning of optimal edge detection is: the algorithm can identify as many actual edges in the image as possible; the identified edges should be as close as possible to the actual image. As close as possible to the actual edge of ; an edge in an image can only be identified once, and possible image noise should not be identified as an edge. Canny uses a variational method, and the Canny algorithm is suitable for different occasions, and its parameters allow adjustments to identify different edge characteristics according to the specific requirements of different implementations.

In the specific embodiment of the present invention, the method of using the canny edge detection operator to display the edge map of the object edge is only a preferred embodiment of the present invention, and is not used to limit the protection scope of the present invention. Other methods that can display the object edge more accurately The method is applicable to all the embodiments of the present invention.

Preferably, step 23 includes:

Step 231, applying the random hough transform method to obtain the first straight line feature in the edge graph.

In the specific embodiment of the present invention, the Hough transform is a parameter estimation technique using the voting principle. Its principle is to use the point-line duality of image space and Hough parameter space to convert the detection problem in image space to parameter space. A straight line is detected by performing simple accumulation statistics in the parameter space, and then finding the peak value of the accumulator in the Hough parameter space. The essence of the Hough transform is to cluster the pixels with a certain relationship in the image space, and to find the cumulative corresponding points in the parameter space that can connect these pixels with a certain analytical form.

This step is based on the straight line feature selection of the random hough transform, and the random hough transform method is used to find the straight line feature in the edge map obtained in step 22, that is, the first straight line feature; because the road edge has obvious straight line features, this step can be Eliminate distracting factors as much as possible.

Further, since the first straight line feature will still have other straight line feature interference except for the road edge, in step 23, it is necessary to perform a second screening in combination with the straight line feature of the road edge of the preset road model to obtain the second straight line features. Specifically, in step 23, the first straight line feature whose similarity with the straight line feature of the road edge of the preset road model is greater than a preset value is determined as the second straight line feature; The first straight line feature whose similarity degree of the edge straight line feature is less than or equal to the preset value is determined as the second straight line feature. Preferably, the calculation of the similarity between two straight line features can be calculated through specific mathematical formulas, such as some functions, etc., which are not described in detail here; and the above-mentioned preset values can be set differently according to different occasions , not limited to a fixed value.

Continuing from the above example, there are errors and uncertainties in any detection, especially in the image processing in the previous steps, interference information must be mixed in individual images, such as green belts, arrows on the road, etc. Objects that will be mistaken for the edge of the road; these falsely detected features are close to the real road edge features, and it is difficult to filter out through step 23. At the same time, due to the continuous operation of the vehicle, in most cases, the position of the road edge in the image is fixed, and the detection result will always jump in a small range around this true value, although the detection result of a single image is not completely reliable , so execute step 24, combine the results of several previous images, average among a certain number of results, you can find a more credible position of the road edge, that is, obtain the predicted road edge information, and then infer the road width and other information.

Then combine the predicted road edge information and the second straight line feature of the current target image to determine the target straight line feature, which is equivalent to roughly predicting the position of the road before detecting the next image, even if the detection result only detects After leaving the left or right edge of the road, the position of the edge on the other side can still be recovered according to information such as the width of the road, and the characteristics of the target straight line can be determined.

Preferably, if the current target image is the first frame image collected, the predicted road edge information cannot be obtained, and the second straight line feature is determined as the target straight line feature; but the road edge information obtained by analyzing the first frame image The reliability of the results is low, and the method of averaging multiple frames of images should be used to improve the accuracy of the results on the road edge.

Further, step 25 is executed to obtain the position of the road edge in the target image according to the target straight line feature, that is, the target straight line feature is the road edge in the target image. Specifically, the road edge refers to two straight lines, and the positions of the four endpoints of the two straight lines in the target image can represent the position of the road edge in the target image.

In the above-mentioned embodiment of the present invention, as shown in Figure 3, step 4 includes:

Step 41, obtaining the actual position of the calibration feature point;

Step 42, according to the position of the calibration feature point on the target image and the actual position, use an inverse perspective transformation method to determine the conversion relationship between the coordinate system of the camera device and the spatial coordinate system;

Step 43: Generate an inverse perspective transformation matrix according to the transformation relationship between the camera coordinate system and the space coordinate system.

In the specific embodiment of the present invention, to figure anti-perspective transformation, only need to transform discrete point column; Inverse perspective transformation is a comparatively mature public algorithm, will not expand here and carry out specific description; Its basic step is to select several point (for example, analyze the image to get the coordinates of the sideline or center line of each row, so as to obtain a series of discrete points), and then measure the actual position of these points (each point has a horizontal and vertical coordinates, the unit is pixel, according to the size of the photosensitive array (several millimeters) transform the coordinates to the actual camera device, and the unit becomes meter), according to the coordinate mapping relationship shown in Figure 4, clarify the conversion relationship between the camera device coordinate system and the space coordinate system, and generate an inverse perspective transformation matrix .

In the above-mentioned embodiment of the present invention, as shown in FIG. 5, step 5 includes:

Step 51, cropping the target image, keeping only the part below the horizon on the target image;

Step 52, performing picture conversion on the cropped target image according to the inverse perspective transformation matrix to obtain a fan-shaped picture;

Step 53: Crop the fan-shaped picture based on the position of the road edge on the target image to obtain a rectangular picture, and the rectangular picture is a front view of the target image.

In the specific embodiment of the present invention, since the front view of the road markings (the picture directly facing downwards) does not contain objects above the horizon, because in order to reduce the calculation amount when converting according to the inverse perspective transformation matrix, the first step is performed in step 51. One crop, keeping only the part below the horizon in the target image. In step 52, the coordinates of each marked point in the image below the horizon are substituted into the inverse perspective transformation formula to obtain the coordinates in the space coordinate system, and a fan-shaped picture is obtained. The fan-shaped picture represents the front view of the target image, and because the road is generally rectangular, a considerable part of the information in the fan-shaped picture has nothing to do with the road, so it is necessary to delete the part that is not related to the road to get the target of attention; Therefore, step 53 is performed, based on the detected lane, appropriately expanded, and the fan-shaped picture is cut for the second time to obtain a rectangular picture, which is the target of interest, that is, the front view of the target image; This method realizes the real-time collection of road markings, and at the same time realizes the automation of calibration, correction, and front view generation, which greatly saves the time for subsequent processing and saves a lot of manpower and material resources.

In order to better achieve the above purpose, as shown in Figure 6, an embodiment of the present invention also provides a road marking collection device based on orthographic projection, including:

Acquisition module 10, is used for utilizing the camera device installed on the collection vehicle to collect road images and obtain target images;

An acquisition module 20, configured to perform image processing on the target image to acquire the position of the road edge in the target image;

An extraction module 30, configured to determine the position of the road according to the position of the road edge in the target image, and extract feature points from the target image based on the position of the road to obtain marked feature points;

The conversion module 40 is used to obtain the conversion relationship between the coordinate system of the camera device and the spatial coordinate system according to the calibration feature points, and generate an inverse perspective transformation matrix;

The generating module 50 is configured to perform picture conversion on the target image according to the inverse perspective transformation matrix to obtain a front view of the target image.

Specifically, the camera device installed on the collection vehicle can be a separate camera, or a camera device embedded in the device, which is similar in principle to the method of embedding a camera in a mobile phone, and in order to obtain the inverse transformation matrix accurately, When the device or camera device is initially installed, the installation height, installation location and its attribute parameters, etc. can be manually or automatically input to provide more accurate data for subsequent calculations.

In the above embodiments of the present invention, the road marking collection device further includes:

The display storage module is used to display and store the front view of the target image in real time.

Further, in the above-mentioned embodiments of the present invention, the acquisition module 20 includes:

The first acquisition submodule is used to convert the target image into a grayscale image;

The second acquisition sub-module is used to convert the grayscale image into an edge image showing only the edge of the object;

The third obtaining sub-module is used to obtain the first straight line feature in the edge map, and filter the first straight line feature according to the straight line feature of the road edge of the preset road model, and obtain the second straight line feature;

The fourth acquisition sub-module is used for if the current target image is the first frame image collected, the second straight line feature is the target straight line feature; otherwise, according to the second straight line of multiple frames of images before the current target image feature, obtaining predicted road edge information, and combining the predicted road edge information and the second straight line feature of the current target image to determine the target straight line feature;

The fifth acquisition sub-module is configured to acquire the position of the road edge in the target image according to the target straight line feature.

Preferably, in the above embodiments of the present invention, the second acquisition submodule includes:

The first conversion module is used to convert the grayscale image into an edge image showing only object edges by using a canny edge detection operator.

Preferably, in the above embodiments of the present invention, the third acquisition submodule includes:

The second conversion module is used to obtain the first straight line feature in the edge graph by applying the random hough transform method.

In the above embodiments of the present invention, the conversion module 40 includes:

The first conversion submodule is used to obtain the actual position of the calibration feature point;

The second conversion sub-module is used to determine the conversion relationship between the coordinate system of the camera device and the spatial coordinate system by using an inverse perspective transformation method according to the position of the calibration feature point on the target image and the actual position;

The third transformation sub-module is used to generate an inverse perspective transformation matrix according to the transformation relationship between the vehicle coordinate system and the space coordinate system.

In the above-mentioned embodiments of the present invention, the generating module 50 includes:

The first generating submodule is used to crop the target image, and only keep the part below the horizon on the target image;

The second generation sub-module is used to perform image conversion on the cropped target image according to the inverse perspective transformation matrix to obtain a fan-shaped image;

The third generation sub-module is configured to cut the fan-shaped picture based on the position of the road edge on the target image to obtain a rectangular picture, and the rectangular picture is a front view of the target image.

In the road marking acquisition method based on orthographic projection in the specific embodiment of the present invention, the road edge is identified by processing the collected target image, and self-calibration is performed through the calculation of the calibration feature point detection and cumulative sampling correction of the road , get the inverse perspective transformation matrix, and then convert the target image into a front view, realize the real-time collection of road markings, and realize the automation of calibration, correction, and front view generation, which greatly saves the time for subsequent processing and saves A lot of manpower and material resources.

It should be noted that all embodiments of the road marking collection method based on orthographic projection described above are applicable to the device of the road marking collection device based on orthographic projection provided by the present invention, and all of them are Can achieve the same or similar beneficial effects.

The above description is a preferred embodiment of the present invention, it should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, some improvements and modifications can also be made, these improvements and modifications It should also be regarded as the protection scope of the present invention.

Claims (14)

1., based on a roadmarking acquisition method for orthogonal projection, it is characterized in that, comprising:

Utilize the camera head collection road image being arranged on and gathering on car, obtain target image;

Image procossing is carried out to described target image, obtains the position of road edge in described target image;

Determine the position of road according to the position of described road edge in described target image, and based on the position of described road, feature point extraction is carried out to described target image, obtain feature point for calibration;

Obtain the transformational relation between camera head coordinate system and space coordinates according to described feature point for calibration, generate inverse perspective mapping matrix;

According to described inverse perspective mapping matrix, picture conversion is carried out to described target image, obtain the front elevation of described target image.

2. the roadmarking acquisition method based on orthogonal projection according to claim 1, is characterized in that, also comprise after obtaining the front elevation of described target image:

The front elevation of described target image is shown in real time and stores.

3. the roadmarking acquisition method based on orthogonal projection according to claim 1, is characterized in that, carry out image procossing to described target image, and the step obtaining the position of road edge in described target image comprises:

Described target image is converted to gray level image;

Described gray level image is converted into the edge figure only showing object edge;

In described edge figure, obtain the first linear feature, and according to the linear feature of the road edge of default road model, described first linear feature is screened, obtain the second linear feature;

If current described target image is the first two field picture gathered, described second linear feature is target line feature; Otherwise, according to the second linear feature of the multiple image before current described target image, obtain predicted link side information, and in conjunction with the second linear feature of described predicted link side information and current described target image, determine target line feature;

According to described target line feature, obtain the position of road edge in described target image.

4. the roadmarking acquisition method based on orthogonal projection according to claim 3, is characterized in that, the step described gray level image being converted into the edge figure only showing object edge comprises:

Canny edge detection operator is utilized described gray level image to be converted into the edge figure only showing object edge.

5. the roadmarking acquisition method based on orthogonal projection according to claim 3, is characterized in that, the step obtaining the first linear feature in described edge figure comprises:

Apply random hough transform method and obtain the first linear feature in described edge figure.

6. the roadmarking acquisition method based on orthogonal projection according to claim 1, is characterized in that, obtains the transformational relation between camera head coordinate system and space coordinates according to described feature point for calibration, and the step generating inverse perspective mapping matrix comprises:

Obtain the physical location of described feature point for calibration;

According to the position of described feature point for calibration on described target image and described physical location, utilize the transformational relation between inverse perspective mapping method determination camera head coordinate system and space coordinates;

According to the transformational relation between described camera head coordinate system and space coordinates, generate inverse perspective mapping matrix.

7. the roadmarking acquisition method based on orthogonal projection according to claim 1, is characterized in that, carry out picture conversion according to described inverse perspective mapping matrix to described target image, the step obtaining the front elevation of described target image comprises:

Described target image is cut, only retains local horizon on described target image with lower part;

According to described inverse perspective mapping matrix, the described target image through cutting being carried out picture conversion, obtaining a fan-shaped picture;

On described target image road edge position based on described fan-shaped picture is cut, obtain a rectangle picture, described rectangle picture is the front elevation of described target image.

8., based on a roadmarking harvester for orthogonal projection, it is characterized in that, comprising:

Acquisition module, for utilizing the camera head collection road image being arranged on and gathering on car, obtains target image;

Acquisition module, for carrying out image procossing to described target image, obtains the position of road edge in described target image;

Extraction module, for determining the position of road according to the position of described road edge in described target image, and carries out feature point extraction based on the position of described road to described target image, obtains feature point for calibration;

Modular converter, for obtaining the transformational relation between camera head coordinate system and space coordinates according to described feature point for calibration, generates inverse perspective mapping matrix;

Generation module, for carrying out picture conversion according to described inverse perspective mapping matrix to described target image, obtains the front elevation of described target image.

9. the roadmarking harvester based on orthogonal projection according to claim 8, is characterized in that, described roadmarking harvester also comprises:

Display memory module, for showing the front elevation of described target image in real time and storing.

10. the roadmarking harvester based on orthogonal projection according to claim 8, it is characterized in that, described acquisition module comprises:

First obtains submodule, for described target image is converted to gray level image;

Second obtains submodule, for described gray level image being converted into the edge figure only showing object edge;

3rd obtains submodule, for obtaining the first linear feature in described edge figure, and screening described first linear feature according to the linear feature of the road edge of default road model, obtaining the second linear feature;

4th obtains submodule, if be the first two field picture gathered for current described target image, described second linear feature is target line feature; Otherwise, according to the second linear feature of the multiple image before current described target image, obtain predicted link side information, and in conjunction with the second linear feature of described predicted link side information and current described target image, determine target line feature;

5th obtains submodule, for according to described target line feature, obtains the position of road edge in described target image.

The 11. roadmarking harvesters based on orthogonal projection according to claim 10, is characterized in that, described second obtains submodule comprises:

First conversion module, is converted into for utilizing canny edge detection operator the edge figure only showing object edge by described gray level image.

The 12. roadmarking harvesters based on orthogonal projection according to claim 10, is characterized in that, the described 3rd obtains submodule comprises:

Second conversion module, obtains the first linear feature for applying random hough transform method in described edge figure.

The 13. roadmarking harvesters based on orthogonal projection according to claim 8, it is characterized in that, described modular converter comprises:

First transform subblock, for obtaining the physical location of described feature point for calibration;

Second transform subblock, for according to the position of described feature point for calibration on described target image and described physical location, utilizes the transformational relation between inverse perspective mapping method determination camera head coordinate system and space coordinates;

3rd transform subblock, for according to the transformational relation between described vehicle axis system and space coordinates, generates inverse perspective mapping matrix.

The 14. roadmarking harvesters based on orthogonal projection according to claim 1, it is characterized in that, described generation module comprises:

First generates submodule, for cutting described target image, only retains local horizon on described target image with lower part;

Second generates submodule, for the described target image through cutting being carried out picture conversion according to described inverse perspective mapping matrix, obtains a fan-shaped picture;

3rd generates submodule, and cut described fan-shaped picture based on the position of road edge on described target image, obtain a rectangle picture, described rectangle picture is the front elevation of described target image.