KR-20260063538-A - Deep Learning-based Geographic Information System Applicable to Small Airport

KR20260063538AKR 20260063538 AKR20260063538 AKR 20260063538AKR-20260063538-A

Abstract

The present invention relates to a deep learning-based navigation safety information system applicable to small airports, characterized by comprising: a camera installed in the direction of the runway to photograph aircraft taking off and landing; a server that receives image data captured by the camera and performs image processing using a deep learning algorithm; and a display that displays the data received from the server on a PC or control center for monitoring.

Inventors

지민석
김동준
백남준

Assignees

한서대학교 산학협력단

Dates

Publication Date: 20260507
Application Date: 20241030

Claims (4)

A camera installed facing the runway to photograph aircraft taking off and landing; A server that receives video data captured by the above camera and performs video processing using a deep learning algorithm; and A deep learning-based navigation safety information system applicable to small airports, characterized by including a display that monitors video data received from the above-mentioned server by displaying it on a PC or control center.
In paragraph 1, The above server is, A collection unit that receives video data captured through a camera and inputs it into an internal server processor or storage unit, and A preprocessing unit that removes noise and normalizes the above-mentioned input image data using a preset filter, and A processing unit that detects and classifies aircraft and detects aircraft identification code areas using artificial intelligence based on the above-mentioned preprocessed image data, and A recognition unit that extracts the above aircraft identification code area and recognizes the aircraft identification code through a deep learning-based OCR model, and A storage unit storing data regarding the above-mentioned aircraft, received image data, preprocessed image data, algorithms and training data used in the processing unit, and A deep learning-based navigation safety information system applicable to small airports, characterized by including a control unit that controls the flow of data between the above-mentioned collection unit, preprocessing unit, processing unit, recognition unit, and storage unit.
In paragraph 2, A deep learning-based navigation safety information system applicable to small airports, characterized by the processing unit constructing a dataset of aircraft types and conducting training through a deep learning-based YOLO V4 model to automatically detect aircraft, thereby detecting aircraft detection and classification and aircraft identification code areas.
In paragraph 2, A deep learning-based navigation safety information system applicable to small airports, characterized in that the recognition unit learns using a labeling method to reduce the recognition rate and error rate of a deep learning-based OCR model, and performs image preprocessing using OpenCV's threshold and contours.

Description

Deep Learning-based Geographic Information System Applicable to Small Airports The present invention relates to a deep learning-based navigation safety information system applicable to small airports, and more specifically, to a deep learning-based navigation safety information system applicable to small airports capable of detecting aircraft and identifying movement paths and locations within a ground runway at small airports. Airport Surface Detecting Equipment (ASDE) detects moving objects within the airport movement area (runways, taxiways, aprons) and serves as an auxiliary device to the Primary Radar as a facility that assists in preventing collisions on the ground and guiding aircraft approaches. Recently, measures to prevent collision accidents and reduce airport congestion are being proposed using advanced analysis systems capable of identifying and predicting the movement paths of aircraft and other moving objects on the runway using ASDE information. However, these radar-based surveillance systems require significant manpower and costs for construction and maintenance, and their use at small airports is limited because they can only detect aircraft exceeding a target cross-section at a specific angle and altitude above the horizontal plane. In addition, Automatic Dependent Surveillance-Broadcasting (ADS-B), which offers high precision and low installation, operation, and maintenance costs, was installed on aircraft to allow nearby users to determine location upon receiving transmissions from transmitters; however, as concerns regarding GPS stability were raised in the early 2000s, it is now recommended to use ADS-B in conjunction with equipment capable of verifying the validity of the data. Nevertheless, there are limitations regarding locations free from radio interference around airports, and it is not applicable to all airports; it is primarily installed at airports with high traffic volumes. FIG. 1 is a block diagram showing a deep learning-based navigation safety information system applicable to small airports according to the present invention. Embodiments of the present invention are described below with reference to the attached drawings so that those skilled in the art can easily implement the invention. However, the present invention may be embodied in various different forms and is not limited to the embodiments described herein. Then, a preferred embodiment of a deep learning-based navigation safety information system applicable to small airports according to the present invention will be described in detail. FIG. 1 is a block diagram showing a deep learning-based navigation safety information system applicable to small airports according to the present invention. Referring to FIG. 1, a deep learning-based navigation safety information system applicable to a small airport according to the present invention may include a camera (100), a server (200), and a display (300). The above camera (100) can be installed in the direction of the runway to photograph aircraft taking off and landing and transmit video data to the server (200). Here, the above cameras (100) are installed in total of three locations to photograph both ends and the center of the runway, the cameras at both ends detect aircraft taking off and landing, and the camera in the center can perform identification code area detection and aircraft classification. The above server (200) can receive image data and perform image processing using a deep learning algorithm. Such a server (200) may include a collection unit (210), a preprocessing unit (220), a processing unit (230), a recognition unit (240), a storage unit (250), and a control unit (260). The above collection unit (210) can receive video data captured through the camera (100) and input it into the internal processor or storage unit (250) of the server (200). The above preprocessing unit (220) can remove noise and normalize the input image data using a preset filter. That is, the image preprocessing unit (220) can remove noise using a digital image filter and normalize the input image through white balancing, color temperature adjustment, angle correction, distortion correction, size correction, blur correction, etc. The above processing unit (230) defines the target object (aircraft) to be found in the preprocessed image data and processes the image data using artificial intelligence to detect and classify the aircraft and detect the aircraft identification code area. That is, the processing unit (230) can detect aircraft and classify aircraft and detect aircraft identification code areas by building a dataset of aircraft types and performing learning through a deep learning-based YOLO V4 model to automatically detect aircraft. The above recognition unit (240) can recognize the identification code of an aircraft. Here, the aircraft identification code consists of letters and numbers, so an OCR algorithm is applied to detect and recognize the letters in a natural backg