Super-resolution reconstruction method, system and device for underwater fish image

Abstract

The embodiment of the invention provides a super-resolution reconstruction method, system and device for an underwater fish image. The method comprises the steps of obtaininga low-resolution fish original image shot by an underwater camera; and inputting the low-resolution fish original image into a generator network in an image super-resolution reconstruction model, and obtaining a super-resolution reconstructed high-resolution underwater fish image corresponding to the low-resolution fish original image. The device comprises an underwater camera, a light source, an illuminance transmitter and a control processor, wherein the illuminance transmitter can sense the light intensity of the environment so as to control the illumination intensity of the light source through the control processor to supplement light for the underwater camera. According to the embodiment of the invention, the method achieves the high-resolution reconstruction of the low-resolution fish original image throughthe construction of the generation and judgment network model, solves a problem that the collected underwater fish image is poor in quality because of the low and non-uniform illuminance of a culturesite and the turbidity of a water body, and provides an accurate, continuous and consistent basis for the detection of small target fishes.

Description

technical field [0001] The invention relates to the technical field of aquatic information, in particular to a method, system and device for super-resolution reconstruction of underwater fish images. Background technique [0002] Super-resolution (SR) reconstruction technology is an image processing technique that reconstructs low-resolution (LR) images into higher-resolution (HR) images. In addition to improving image perception quality, it also helps to improve applications such as object detection Effect. Due to the low and uneven illumination of the aquaculture site, the turbidity of the water body and the limitation of camera equipment, the quality of the collected underwater fish images is poor. By using the method of image super-resolution reconstruction, it is beneficial to restore the lost high-frequency feature information of the collected fish images, so that the subsequent images used for small target fish recognition have more target information features, and i...

AI Technical Summary

Problems solved by technology

However, as the number of layers increases, the convergence of the model becomes more difficult, the effect of the model becomes more unstable, the brightness of the reconstructed image is inconsistent, and it also leads to insufficient utilization of shallow layer information

Although BN accelerates the network convergence speed and can use a larger learning rate, it introduces noise that causes the reconstructed picture to be accompanied by artifacts, and the model performance is unstable, making it difficult to apply to super-resolution reconstruction of underwater fish images.

[0005] Based on the many defects in the ultra-clear reconstruction of underwater fish images in the existing technology, it is urgent to provide a new image processing method. On the basis of strengthening the monitoring of image brightness after reconstruction and the stability of the model structure, not only Produces better image reconstructions and also avoids unrealistic textures

Images