Moving ship detection method and device for low-resolution time sequence remote sensing image

Abstract

The invention discloses a moving ship detection method and device for low-resolution time sequence remote sensing images. The detection method comprises the following steps: acquiring low-resolution remote sensing images of different time sequences; splicing two remote sensing images with different time sequences and mapping the two remote sensing images to an RGB color space to synthesize a false color image; performing quantization processing on the false color image, and labeling the ship target to obtain a labeled image training set and a labeled image verification set; and carrying out feature extraction on a to-be-detected low-resolution time sequence remote sensing image by utilizing a pre-constructed ship detection model based on the improved YOLOv4 network, and labeling a ship target. According to the invention, the time sequence features and the spatial features of the remote sensing image can be extracted, the detection precision of the ship target is improved, and the probability of missing detection and false alarm is reduced.

Description

technical field [0001] The invention relates to the technical field of computer image processing, and more specifically relates to a moving ship detection method and device for low-resolution time-series remote sensing images. Background technique [0002] Low-resolution remote sensing image ship detection technology has broad prospects in ocean monitoring, and can be effectively applied to many aspects such as port dynamic monitoring, naval battlefield situation awareness, sea area security, and military target detection, so it has great research value . [0003] At present, the popular deep learning method has been able to realize ship detection in medium and high-resolution remote sensing images, but the imaging quality of low-resolution remote sensing images is poor, the spatial resolution is low, and the interference of broken clouds and thick clouds is serious, which leads to the detection of ships. A large number of missed detections and false alarms are prone to occ...

AI Technical Summary

Problems solved by technology

[0003] At present, the popular deep learning method has been able to realize ship detection in medium and high-resolution remote sensing images, but the imaging quality of low-resolution remote sensing images is poor, the spatial resolution is low, and the interference of broken clouds and thick clouds is serious, which leads to the detection of ships. A large number of missed detections and false alarms are prone to occur, and it is difficult to improve the detection accuracy

Images