Image super-resolution reconstruction model training method, ship tracking method and device
A model training and super-resolution technology, applied in the field of computer vision, can solve the problem of low accuracy of ship recognition, achieve reconstruction accuracy, facilitate tracking, and improve accuracy
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0029] An embodiment of the present invention provides an image super-resolution reconstruction model training method, which is applied to ship image recognition, such as figure 1 shown, including the following steps:
[0030] S11: Acquire ship image training samples, where the ship image training samples are high-resolution training samples.
[0031] Exemplarily, the image super-resolution reconstruction model learns the distribution relationship between low-resolution images and high-resolution images. In order to improve the recognition accuracy, in the embodiment of the present invention, the ship image training samples should ensure image diversity, including different shooting environment, different sensors, different degraded models, etc., the ship image training sample can be collected and processed by a certain camera and stored in the terminal, directly called from the terminal, or obtained from a network database. The embodiment does not limit the acquisition metho...
Embodiment 2
[0057] An embodiment of the present invention provides a ship tracking method, such as figure 2 shown, including the following steps:
[0058] S21: Acquiring video images collected by video satellites.
[0059] Exemplarily, the video image collected by the video satellite has low spatial resolution and the image is blurred. The acquisition method may be to deframe the video collected by the video satellite to obtain each frame of video image; Video images of a certain number of frames are used as acquired video images, for example, a video image is acquired every 3 frames of video. This embodiment does not limit the manner of acquiring video images collected by the drone, which can be determined according to requirements.
[0060] S22: Input the video image into the image super-resolution reconstruction model to obtain the ship detection image; the image super-resolution reconstruction model is generated by training the image super-resolution reconstruction model training m...
Embodiment 3
[0067] The embodiment of the present invention also provides an image super-resolution reconstruction model training device, which is applied to ship image recognition, such as image 3 shown, including:
[0068] The first acquiring module 31 is configured to acquire training samples of ship images, which are high-resolution training samples; see step S11 in Embodiment 1 for specific implementation, and will not be repeated here.
[0069] The low-resolution training sample acquisition module 32 is configured to obtain low-resolution training samples according to the high-resolution training samples; see step S12 in Embodiment 1 for the specific implementation, and will not be repeated here.
[0070] The first decomposing module 33 is used to decompose the low-resolution training samples into sparse representation according to the low-resolution sparse representation algorithm to obtain the low-resolution sparse training samples; see step S13 in Embodiment 1 for the specific im...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More