Target Ship Detection and Tracking Method Based on Optical Remote Sensing Image

Abstract

The present invention provides a ship detection and tracking method in a thin cloud scene based on optical remote sensing images. The detection method includes: obtaining the optical remote sensing image sequence corresponding to the ship; cutting the optical remote sensing image sequence according to the size of the area of ​​interest, and then image Enhancement, followed by median filtering to obtain the grayscale image sequence of the ship's wake; normalize the grayscale image sequence, and then perform grayscale peak detection to obtain the grayscale peak detection result; perform grayscale peak detection results Threshold detection to obtain the coordinates of the ship wake in the grayscale image sequence. On the basis of the detection, the coordinates of the ship's wake in the gray-scale image sequence are further associated with data to obtain the track of the ship in the gray-scale image sequence, and the target ship is tracked according to the track to realize the tracking of the target ship. Ship tracking with high accuracy and high success rate. The detection and tracking method of the target ship has low computational complexity, fast operation speed, simple algorithm and easy engineering implementation.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a target ship detection and tracking method based on optical remote sensing images. Background technique [0002] The orbital period of the geostationary optical remote sensing satellite is the same as the earth's rotation period, and it remains relatively stationary with the earth. It has a large imaging width, and can continuously observe a large area of ​​designated regions of interest. It has the advantages of high time resolution and good timeliness of data. For example, the Gaofen-4 satellite is the earth observation satellite with the highest geosynchronous orbit resolution in the world today. It was launched in 2015 and is located above the equator at 105.6° east longitude, about 36,000km above the ground. The spatial resolution of sub-satellite points is better than 50m×50m, sub-satellite point single-scene imaging coverage area is close to 500km×500km, and can carry out...

AI Technical Summary

Problems solved by technology

However, because the geometry and texture features of ship wakes will change with the change of motion speed, the target detection method based on geometry and texture features is not applicable; because the average gray value of the image will change with the change of cloud cover , so the target detection method based on the gray threshold is not applicable; since a large number of samples are required to train the estimator, and there is no such training sample library, the target detection method based on machine learning and deep learning is also not applicable

Images