Weak and small airspace target detection method based on super-resolution feature enhancement

Abstract

The invention discloses a weak and small airspace target detection method based on super-resolution feature enhancement. The method comprises the following steps: 1, performing illumination compensation on an airspace video stream to obtain an airspace video image; 2, acquiring a foreground target gray feature map by adopting a foreground detection algorithm based on background modeling and inter-frame difference, extracting a foreground target contour by morphological filtering, and calculating to obtain a prediction frame set of a foreground target; 3, adopting a multi-target tracker based on a KM algorithm and Kalman filtering to track the prediction frames, and screening out suspected target frames; 4, training a super-resolution neural network model on the spatial domain target super-resolution data set, and performing super-resolution enhancement on a foreground image region mapped by the suspected target frame by using the super-resolution model in a detection stage; and 5, training a neural network model based on a YOLOv4 framework on the airspace target detection data set, and detecting a target in the foreground target area after super-resolution enhancement by using a detector model in a detection stage to obtain a target detection result.

Description

technical field [0001] The invention relates to the field of computer vision, in particular to a method for detecting weak and small airspace targets based on super-resolution feature enhancement. Background technique [0002] In recent years, with the gradual opening of low-altitude airspace and the development of related technologies, drones have been widely used in aerial photography, agriculture and forestry, logistics, security and other industries. At the same time, with the prosperity and development of the drone market and the blowout of the number of drones, while drones bring convenience to life and production, they also bring many security and privacy risks. Small UAVs have the characteristics of high mobility and high concealment, and there are not many operators with professional flying skills and literacy. Incidents of UAVs flying in dangerous areas without reporting or approval occur frequently, which poses a threat to the safety of public places. , aviation ...

AI Technical Summary

Problems solved by technology

However, the image area of ​​long-distance drones is small and the features are not obvious. There are many challenges in the current weak airspace target detection problem. The existing neural network target detection method cannot directly detect the target. It is urgent to design an accurate, reliable, low- Weak and small airspace object detection method based on missed detection rate

Images