Ship Trajectory Prediction via Heading-Aware Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional ship detection methods based on horizontal rectangular frames are inefficient due to excessive background information inclusion, incorrect handling of overlapping ships, strong frame overlapping, and failure to account for ship heading changes, leading to inaccurate tracking and trajectory prediction.
Innovation Solution
A ship image trajectory tracking and prediction method based on ship heading recognition, which involves preprocessing ship image data sets, training a rotating ship detection network, tracking ship trajectories using ship heading information, and predicting future navigation trajectories and headings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If horizontal rectangular frames are used for ship detection, then the detection process is simple, but the detection accuracy deteriorates due to excessive background information inclusion
Solution Approach 1:
The patent segments the ship detection problem by dividing it into two stages: first detecting ships with horizontal rectangular frames, then refining the detection with rotating rectangular frames. This segmentation allows the system to benefit from both the simplicity of horizontal frames and the accuracy of rotating frames, resolving the contradiction between detection process simplicity and detection accuracy.
Solution Approach 2:
The patent introduces rotation angle as an additional dimension to the traditional horizontal rectangular frame detection. By adding the rotation dimension, the detection framework transitions from 4 parameters (x, y, w, h) to 5 parameters (x, y, w, h, θ), enabling the detection of ships at various orientations and significantly improving detection accuracy without excessively complicating the process.
2Productivity
If horizontal rectangular frames are used for ship detection, then the algorithm is computationally efficient, but tracking accuracy deteriorates due to failure to account for ship heading changes
Solution Approach 1:
The patent performs preliminary ship detection using computationally efficient horizontal rectangular frames to obtain initial ship positions and headings. This preliminary action provides a good starting point for subsequent tracking, allowing the system to maintain computational efficiency while still capturing essential ship orientation information needed for accurate tracking.
Solution Approach 2:
The patent implements a feedback mechanism where ship heading information detected in one frame is used to guide the search and tracking in subsequent frames. The detected ship headings provide feedback that helps maintain tracking accuracy over time, allowing the system to adapt to ship orientation changes while preserving computational efficiency through informed search regions.
3Speed
If horizontal rectangular frames are used, then the detection algorithm is fast, but trajectory prediction accuracy deteriorates due to inability to capture ship orientation
Solution Approach 1:
The patent extracts ship heading information from the detection process by incorporating rotation angle estimation into the detection framework. By taking out and explicitly modeling the orientation component, the system gains the ability to predict ship trajectories more accurately while maintaining relatively fast detection speeds through efficient rotation angle calculation methods.
Data Source
AI summary
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.