A tower flight command training scene generation method based on a training flight schedule

Abstract

The application provides a tower flight command training scene generation method based on a training flight schedule, which can replace most of the manual input training scene process and improve the efficiency of producing flight command training scenes. It comprises the following steps: step one, obtaining single cell image data in the training flight schedule; step two, obtaining single flight plan line segment position information; step three, screening the information required for generating a training scene according to the single cell image data obtained in step one and the single flight plan line segment position information obtained in step two, and obtaining a screenshot list of required text information; step four, removing interference and error details according to the screenshot list of required text information obtained in step three, and obtaining a processed text information screenshot list; and step five, traversing each screenshot to extract the text therein and saving the text in corresponding line segment data information through a Tesseract character recognition framework according to the processed text information screenshot list obtained in step four.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of air traffic control teaching training, and relates to a tower flight command training scene generation method based on a training flight schedule. BACKGROUND

[0002] The tower command area of the training airport includes ground operation, maneuvering flight within the visual range of the tower, and visual approach to and from the port. The tower commander is mainly responsible for: controlling the aircraft's start, taxi, takeoff, landing, takeoff and landing route, and related maneuvering flight, providing takeoff and landing conditions and flight dynamic information services to the crew, assisting the crew in handling abnormal and emergency situations, coordinating and transferring with the approach control according to the agreement. Allocate the takeoff and landing route, visual maneuvering, and visual joining and spacing of the aircraft, effectively monitor and remind the aircraft below 100m from the ground, real-time master the changes of the weather conditions of the airport and the activities affecting the flight, timely and effective monitoring of the ground start and taxi aircraft, correct command and adjustment of the aircraft with special situations, and technical command when the single flight and student crew training show abnormal trends.

[0003] At present, the image recognition technology is developing rapidly, and the open source image processing and recognition framework is mature. The recognition accuracy of the image recognition method provided by the open source image processing and recognition framework can meet the needs of multiple air traffic control use scenarios. However, the schedule generation in the existing tower flight command training system mainly relies on the manual input of the paper schedule by the instructors, the input form is relatively single, and the speed is slow. SUMMARY

[0004] In view of the above defects, the application provides a tower flight command training scene generation method based on a training flight schedule, which can replace most of the manual input of the training scene process and improve the efficiency of producing the flight command training scene.

[0005] The application is implemented through the following technical solutions.

[0006] A tower flight command training scene generation method based on a training flight schedule, comprising:

[0007] Step 1: Obtain single cell image data in the training flight schedule;

[0008] Step 2: Obtain single flight plan line segment position information;

[0009] Step 3: According to the single cell image data obtained in step 1 and the single flight plan line segment position information obtained in step 2, screen the information required for generating the training scene, and obtain a screenshot list of the required text information;

[0010] Step four, according to the screenshot list of the required text information obtained in step three, remove interference and error details to obtain a processed text information screenshot list;

[0011] Step five, according to the processed text information screenshot list obtained in step four, traverse each screenshot to extract the text therein by a Tesseract character recognition framework and save it in the corresponding line segment data information.

[0012] Advantages of the present application:

[0013] 1. Compared with the prior art, the present application fully utilizes the image processing framework and the character recognition framework to perform image recognition on a given schedule, extract schedule information, and then generate a scene to improve teaching efficiency, while ensuring that the generated scene meets the training scene generation requirements;

[0014] 2. In order to eliminate the situation that the length of the line segment does not match the actual length due to unclear line segment images during the scanning process, the present application traverses the start point and end point of the line segment in the line segment list, calculates the length of the line segment, and compares it with the preset line segment length to remove the line segments in the line segment list that do not meet the length requirement.

[0015] 3. In the character recognition process, Tesseract's recognition of characters may be interfered by dashed lines, straight lines or other color spots to some extent. The present application uses a dashed line removal and color filtering method to remove possible interference details. DETAILED DESCRIPTION

[0016] The present application will be further described in detail below.

[0017] One of the tower flight command training scene generation methods based on a training flight schedule of the present embodiment specifically includes the following steps:

[0018] Step one, obtain single cell image data in the training flight schedule; specifically, the following method is used:

[0019] First, perform grayscale processing on the schedule scanning image, then use the Hough line detection method provided by OpenCV to recognize the line segments in the table; traverse these intersection coordinates in groups of four (i.e. top left, bottom left, top right, bottom right), cut the image data of a single cell, and save it in the program execution path in PNG image format.

[0020] Step two, obtain single flight plan line segment position information; the specific steps are as follows:

[0021] 2.1 According to the preset line segment color and error requirement, generate a filter for RGB value range, traverse the pixels of the flight plan picture in order from top left to bottom right, and obtain the position information of all points that match the preset line segment color range, and store them in the complete matching pixel point list;

[0022] 2.2 Traverse all the pixels in the 3x3 range around each pixel point in the complete matching pixel point list in turn, and store the points within the preset color range in the left and right adjacent lists according to the relative position, while storing the number of left adjacent points and the number of right adjacent points in the structure of the pixel point;

[0023] 2.3 Traverse the picture in order from top left to bottom right, obtain the first pixel point that matches the required line segment color as the process point; initialize a line segment list to save the information of each line segment;

[0024] 2.4 Traverse the left adjacent point of the left adjacent point of the process point, obtain the left adjacent point with the most left adjacent points by comparison, set it as the new process point, repeat the above operation until there is no adjacent point on the left side of the process point, and set the starting point of the line segment as the process point;

[0025] 2.5 Starting from the starting point of the line segment, obtain the right adjacent point in the same way as obtaining the left adjacent point, until the rightmost side of the line segment, at which time the end point is equal to the process point, and in the traversal process, all process points are recorded in the line segment structure;

[0026] 2.6 Obtain the straight line equation of the line segment through the starting point and end point coordinates of the line segment, traverse the coordinates of each point in the line segment structure, check whether these coordinates conform to the straight line equation of the line segment, if not, discard the line segment, if yes, keep the line segment. And store it in the line segment list;

[0027] 2.7 Continue to traverse the remaining line segments in the image, if the obtained point coordinates are located in the rectangle formed by the starting point and end point of the previous line segment, discard the point and continue to traverse, if not overlapping with the previous line segment, repeat the process of obtaining all the next line segments, until all the required line segments in the image are extracted and stored in the line segment list.

[0028] Through this method, if the preset line segment color is red, this step will filter out the black line segments that constitute the table, as well as the blue flight plan line segments and other colors except red.

[0029] In order to eliminate the scanning process caused by the length of the line segment image is not clear that the length of the line segment does not match the actual situation, this step by traversing the line segment list of line segment start and end point, calculate the length of the line segment and the preset line segment length compared to remove line segment list of line segment does not meet the length requirements.

[0030] Step three, according to the single cell image data obtained in step one and the single flight plan line segment position information obtained in step two, the information required to generate the training scene is screened, and the screenshot list of the required text information is obtained.

[0031] The implementation principle of this step is that the plan book scanning image contains the information required to generate the training scene, including the time of take-off and landing, the pilot's code, the call sign, the airspace number, the formation number, the flight height and other information. These information are presented in the form of text in the image, and are surrounded around the flight plan line segment in a relatively fixed position. By traversing the line segment position recorded in the line segment list obtained in step two, the required text information screenshot can be selected in the image data list obtained in step one, and stored in the required text information screenshot list. The list contains the mapping relationship between the screenshot and the flight plan line segment.

[0032] Step four, according to the screenshot list of the required text information obtained in step three, remove the interference and error details, and obtain the processed text information screenshot list.

[0033] In the existing technology, the recognition of characters by Tesseract will be interfered by the dotted line, straight line or other color spots to some extent. Therefore, the dotted line removal and color filtering method is adopted in this embodiment to remove the details that may cause interference. The specific steps are as follows:

[0034] 4.1 Traverse each column of the single text image from left to right. When the arrangement of colored pixels and white points in each column meets the fixed proportional relationship, it is determined that the column contains vertically distributed dotted lines or straight lines. The image to the left of the left two pixel points of the dotted line or straight line and the image to the right of the right two pixel points of the dotted line or straight line are truncated and spliced to remove the dotted line or straight line. Repeat the process until all the vertically distributed dotted lines or straight lines are completely removed.

[0035] 4.2 Traverse each row of the single text image from top to bottom. When the arrangement of colored pixels and white points in each row meets the fixed proportional relationship, it is determined that the row contains horizontally distributed dotted lines or straight lines. The image above the upper one pixel point of the dotted line or straight line and the image below the lower one pixel point of the dotted line or straight line are truncated and spliced to remove the dotted line or straight line. Repeat the process until all the horizontally distributed dotted lines or straight lines are completely removed.

[0036] 4.3 Traverse the pixel points in the image from the top left to the bottom right, compare the color of each pixel point with the RGB range of the preset color, retain all the pixel points with matching colors, and replace the points with RGB values not in the RGB range of the preset color with white, so as to remove the interference colors other than the required color;

[0037] 4.4 Store the screenshots of the processed text information after the above process in a processed text information screenshot list.

[0038] Step five, according to the processed text information screenshot list obtained in step four, traverse each screenshot through the Tesseract character recognition framework to extract the text therein and save it in the corresponding line segment data information.

[0039] In specific implementation, the text contains the information of take-off and landing time, pilot code, call sign, airspace number, formation number and flight height; wherein:

[0040] The filter item is set to be a number for the take-off and landing time, call sign, formation number and flight height, and all the text with non-number recognition result is removed;

[0041] The filter item is set to be Chinese for the pilot code, and other types of text are removed;

[0042] The filter item is set to be English and number for the airspace number, and other types of text are removed.

[0043] Then the processed text is saved in the data structure corresponding to each line segment, and the whole flight plan information extraction process is completed.

[0044] Although the embodiments of the present application are described, several improvements can be made by those skilled in the art without departing from the principles of the present application, which should also be considered as belonging to the protection scope of the present application.

Claims

1. A tower flight direction training scene generation method based on a training flight schedule, characterized in that, The method comprises the following steps: Step one, obtaining single cell image data in the training flight plan table; Step two, obtaining single flight plan line segment position information; Step three, screening the information required for generating a training scene according to the single cell image data obtained in step one and the single flight plan line segment position information obtained in step two, and obtaining a screenshot list of required text information; Step four, removing interference and error details according to the screenshot list of required text information obtained in step three, and obtaining a processed text information screenshot list; Step five, traversing each screenshot to extract the text therein and saving the text in corresponding line segment data information by using a Tesseract character recognition framework according to the processed text information screenshot list obtained in step four; The specific steps for obtaining single flight plan line segment position information are as follows: 2.1 Generating a filter RGB value range according to a preset line segment color and error requirement, traversing the pixels of the plan book picture in the order from top left to bottom right, and obtaining the position information of all the points that match the preset line segment color range and storing the information in a completely matched pixel point list; 2.2 Traversing all the pixel points in the 3*3 range around each pixel point in the completely matched pixel point list in turn, and storing the points within the preset color range in the left adjacent list and the right adjacent list of the pixel point according to the relative position, while storing the number of left adjacent points and the number of right adjacent points of the pixel point in the structure body of the pixel point; 2.3 Traversing the picture in the order from top left to bottom right, obtaining the first pixel point that matches the required line segment color as a process point, and initializing a line segment list to save the information of each line segment; 2.4 Traversing the left adjacent point of the left adjacent point of the process point, obtaining the left adjacent point with the most left adjacent points by comparison, setting the left adjacent point as a new process point, and repeating the above operation until there is no adjacent point on the left side of the process point, and setting the starting point of the line segment as the process point; 2.5 Starting from the starting point of the line segment, obtaining the right adjacent points in the same way as obtaining the left adjacent points, until the rightmost point of the line segment, at which time the ending point is equal to the process point, and recording all the process points in the line segment structure body in the process of traversing; 2.6 Obtaining the straight line equation of the line segment through the starting point and the ending point coordinates of the line segment, traversing the coordinates of each point in the line segment structure body, checking whether the coordinates conform to the straight line equation of the line segment, discarding the line segment if not, and retaining the line segment and storing the line segment in the line segment list if yes; 2.7 Continue to traverse the remaining line segments in the image, discard the point if the coordinates of the point are located in the rectangle formed by the starting point and the ending point of the previous line segment, and continue to traverse if the point does not overlap with the previous line segment, repeat all the processes of obtaining the next line segment until all the line segments that meet the requirements in the image are extracted and stored in the line segment list.

2. The method of claim 1, wherein, The single cell image data in the training flight plan table is obtained in the following manner: Firstly, the schedule scanning image is processed in grayscale, and then the Hough line detection method provided by OpenCV is used to identify the line segments in the table; the intersection coordinates of the horizontal and vertical line segments are traversed in groups of four, the image data of a single cell is cut and saved in PNG image format under the program execution path.

3. The method of claim 1 or 2, wherein, The interference details are removed by using the dashed line removal and color filtering method, and the specific steps are as follows: 4.1 Traverse each column of a single text image from left to right. When the arrangement of colored pixels and white pixels in each column meets the fixed proportion relationship, it is determined that the column contains vertically distributed dashed lines or straight lines. The image to the left of the left two pixels of the dashed line or straight line and the image to the right of the right two pixels of the dashed line or straight line are spliced to remove the dashed line or straight line. Repeat the process until all the vertically distributed dashed lines or straight lines are completely removed; 4.2 Traverse each row of a single text image from top to bottom. When the arrangement of colored pixels and white pixels in each row meets the fixed proportion relationship, it is determined that the row contains horizontally distributed dashed lines or straight lines. The image above the upper pixel of the dashed line or straight line and the image below the lower pixel of the dashed line or straight line are spliced to remove the dashed line or straight line. Repeat the process until all the horizontally distributed dashed lines or straight lines are completely removed; 4.3 Traverse the pixels in the image from the top left to the bottom right. Compare the color of each pixel with the RGB range of the preset color. While retaining all the pixels of the matching color, replace the pixels with RGB values outside the RGB range of the preset color with white, thereby removing the interference colors other than the required color; 4.4 Store the screenshots of the processed text information in the processed text information screenshot list.

4. The method according to claim 1 or 2, wherein: The text in step five contains information such as take-off and landing time, pilot code, call sign, airspace number, formation number and flight height; For take-off and landing time, call sign, formation number and flight height, set the filter item to be numbers, and remove all text recognition results that are not numbers; For pilot code, set the filter item to be Chinese, and remove other types of text; For airspace number, set the filter item to be English and numbers, and remove other types of text.