A Fast Automatic Target Detection Method

Abstract

The invention mainly belongs to the technical field of target intrusion detection, and in particular relates to a target detection method for area intrusion based on unmanned aerial vehicle images. The method is used for the target detection of the unmanned aerial vehicle, and the original image in the original video obtained by the onboard camera of the unmanned aerial vehicle is carried out to Gaussian pyramid layering, to reduce the computational complexity of feature point extraction; then extract the image SIFT Image registration is performed on feature points, and the motion information in the image is captured by the LK sparse optical flow of the pyramid to realize the motion calculation of the target point, the clustering of the motion point and the elimination of false targets, and finally the target judgment to realize the target detection. The method can reduce the search range of inter-frame feature points of the UAV video, overcome the problem of large motion displacement in the UAV image, and improve the detection ability; thereby reducing the intensity of manual detection of area intrusion and improving the automatic perception of the UAV ability.

Description

technical field [0001] The invention mainly belongs to the technical field of target intrusion detection, and in particular relates to a target detection method for area intrusion based on unmanned aerial vehicle images. Background technique [0002] Target detection is the operation of separating the target of interest from the background area of ​​a single frame image or sequence image in a surveillance scene, and identifying and extracting meaningful object entities from the image. The prerequisite for UAVs to complete various tasks is to quickly and accurately detect targets in the surveillance scene. The current research on UAV moving target detection algorithms is in the stage of designing specific methods for specific problems, and the adaptability to complex and changeable working scenarios is poor. Also, the level of object detection required varies depending on the application environment. Generally speaking, the primary task of target detection is to search a ce...

AI Technical Summary

Problems solved by technology

For the video images obtained by the payload carried on the UAV, the continuity condition of optical flow is often not satisfied in time domain and air domain

Feature matching can only obtain the optical flow corresponding to the feature points in the local area, and cannot describe the overall motion displacement of the UAV video frame, and the calculation time is long

Images