Ship image trajectory tracking and prediction method based on ship heading recognition

Patent metadata

Field	Value
Publication number	US-12205301-B2
Application number	US-202217886901-A
Country	US
Kind code	B2
Filing date	Aug 12, 2022
Priority date	Jul 6, 2022
Publication date	Jan 21, 2025
Grant date	Jan 21, 2025

Abstract

Disclosed is a ship image track tracking and prediction method based on ship heading recognition, which includes the following steps: obtaining a ship image data set, preprocessing the data set to obtain a preprocessed data set; inputting the preprocessed data set into the rotating ship detection network for training, obtaining the trained rotating ship detection network, collecting the ship navigation video, and inputting the ship navigation video into the trained rotating ship detection network to obtain the ship detection result; inputting the ship detection result into the rotating ship tracking network and tracking the target ship to obtain the historical trajectory and the heading information of the target ship; inputting the historical trajectory and ship heading information of the target ship into the ship trajectory and ship heading prediction network, and predicting the navigation trajectory and ship heading at sea.

First claim

What is claimed is:

1. A ship image trajectory tracking and prediction method based on ship heading recognition, comprising:
S1: obtaining a ship image data set, and preprocessing the ship image data set to obtain a preprocessed data set;
S2: inputting the preprocessed data set into a rotating ship detection network for training to obtain a trained rotating ship detection network, collecting a ship navigation video, and inputting the ship navigation video into the trained rotating ship detection network to obtain a ship detection result;
S3: inputting the ship detection result into a rotating ship tracking network, and tracking a target ship, so as to obtain historical trajectory and ship heading information of the target ship; and
S4: inputting the historical trajectory and the ship heading information of the target ship into a ship trajectory and ship heading prediction network, and predicting the navigation trajectory and ship heading of sea ships.

2. The ship image trajectory tracking and prediction method according to claim 1 , wherein preprocessing the ship image data set comprises:
labeling label data containing the ship heading information by using an 8-parameter rotating frame labeling method, and preprocessing labeled images and labels, and processing the ship heading information by a way of circular smooth label and outputting ship heading data; wherein the label data comprises an abscissa of a center of a marking frame, an ordinate of the center of the marking frame, a width of the marking frame, a height of the marking frame and processed ship heading information.

3. The ship image trajectory tracking and prediction method according to claim 1 , wherein the process of inputting the preprocessed data set into the rotating ship detection network for training comprises:
inputting the preprocessed data set into a backbone network added with a coordinate attention module for ship feature extraction, performing a feature fusion through a feature pyramid, and outputting a first ship feature map; inputting the first ship feature map into a pixel aggregation network for a further ship feature processing, outputting a second feature map; fusing the second feature map based on an adaptive fusion method to obtain a third ship feature map; performing a border and rotation angle regression operation on the third ship feature map, and calculating ship category loss, border loss, rotation angle loss and score confidence loss; minimizing a loss function by using a gradient descent algorithm and updating network parameters to complete the training of the rotating ship detection network.

4. The ship image trajectory tracking and prediction method according to claim 3 , wherein the coordinate attention module aggregates ship feature images input in vertical and horizontal directions into two independent directional perception feature maps through two one-dimensional global pooling operations.

5. The ship image trajectory tracking and prediction method according to claim 3 , wherein the loss function comprises target classification loss, θ classification loss, border regression loss and confidence loss, and the expression of the loss function total is as follows:
total = cls + θ + IOU + obj
wherein IOU is target border regression loss, and cls , θ and obj are the target classification loss, the θ classification loss and confidence loss, respectively.

Assignees

• Univ Shanghai Maritime
• Univ Wuhan Tech

Inventors

• Chen Xinqiang
• Wu Hao
• Yang Yongsheng
• Wu Bing
• Sun Yang
• Wu Huafeng
• Liu Wei
• Xian Jiangfeng

Classifications

• G06T2207/30241

  Trajectory · CPC title

• G06T2207/20081

  Training; Learning · CPC title

• G06T2207/10016

  Video; Image sequence · CPC title

• G01C21/203

  specially adapted for water-borne vessels · CPC title

• G08G3/00

  Traffic control systems for marine craft (marking of navigational route B63B22/16, B63B51/00) · CPC title