Satellite image ship component detection method based on key point regression

Abstract

The invention provides a satellite image ship component detection method based on key point regression, and belongs to the field of satellite image recognizing.The method comprises the steps that firstly, a training data set is constructed, and the ship position, the ship category and the types and positions of components such as a ship gun, a vertical radiation system, a ship bridge, a parking apron, a shipboard number and a take-off and landing runway in a remote sensing image are manually marked; secondly, carrying out data amplification and equalization processing on the training data by adopting a variable coefficient minimization method; then, carrying out target detection on ship targets which are distributed in a satellite image in a random rotation manner by utilizing a ship target and component detection model, and carrying out target slice alignment operation on the feature map and the input image according to a detection result; and finally, carrying out key point coordinate and size regression calculation on the target and the feature map slice to obtain the position, boundary and category of the ship component. Compared with a conventional ship target detection method, the ship target detection method has the advantages that a finer-grained component-level detection result can be obtained, and the ship fine identification capability is realized.

Description

technical field

[0001] The invention belongs to the field of satellite image recognition, and more particularly relates to a satellite image ship component detection method based on key point regression. Background technique

[0002] Ships are an important carrier of maritime transportation, and are of great significance to military reconnaissance, environmental protection, marine detection and other fields. As an important technology of maritime situational awareness, ship target detection is the premise of ship target tracking and state estimation, and has broad application prospects in military and civilian fields. With the rapid development of remote sensing technology, the use of satellite images for ship target detection has the advantages of all-day, all-weather, and no airspace restrictions. However, compared with natural scene images, satellite images have the characteristics of large image size, insufficient resolution, poor imaging quality, target rotation change...

Examples

Embodiment Construction

[0040] At present, some ship target detection methods based on deep learning technology have been improved in data enhancement, feature extraction, feature reuse, loss function design, etc., which alleviates the problems of insufficient resolution of satellite images and small and few targets to a certain extent. However, the detection accuracy of ship targets still needs to be improved. In addition, the existing detection methods can only detect the ship target attribute and position distribution, and cannot determine the key points of the ship and the key components of the ship. Therefore, the present invention proposes a satellite image ship component detection method based on key point regression. The specific flow chart is as follows: figure 2 shown. Examples of satellite imagery ship targets and component annotations are: figure 1 shown.

[0041] The specific embodiments and basic principles of the present invention will be further described below with reference to t...

technical field

[0001] The invention belongs to the field of satellite image recognition, and more particularly relates to a satellite image ship component detection method based on key point regression. Background technique

[0002] Ships are an important carrier of maritime transportation, and are of great significance to military reconnaissance, environmental protection, marine detection and other fields. As an important technology of maritime situational awareness, ship target detection is the premise of ship target tracking and state estimation, and has broad application prospects in military and civilian fields. With the rapid development of remote sensing technology, the use of satellite images for ship target detection has the advantages of all-day, all-weather, and no airspace restrictions. However, compared with natural scene images, satellite images have the characteristics of large image size, insufficient resolution, poor imaging quality, target rotation change...

Examples

Embodiment Construction

[0040] At present, some ship target detection methods based on deep learning technology have been improved in data enhancement, feature extraction, feature reuse, loss function design, etc., which alleviates the problems of insufficient resolution of satellite images and small and few targets to a certain extent. However, the detection accuracy of ship targets still needs to be improved. In addition, the existing detection methods can only detect the ship target attribute and position distribution, and cannot determine the key points of the ship and the key components of the ship. Therefore, the present invention proposes a satellite image ship component detection method based on key point regression. The specific flow chart is as follows: figure 2 shown. Examples of satellite imagery ship targets and component annotations are: figure 1 shown.

[0041] The specific embodiments and basic principles of the present invention will be further described below with reference to t...

Images