Image rotation target detection method based on multistage fusion and angular point offset

Abstract

The invention provides a remote sensing image rotation target detection method based on multistage fusion and angular point offset. The method is used for solving the technical problems that in the prior art, the detection accuracy of targets with different scales is low, and the running speed of the detection process is low. The method comprises the following implementation steps of: 1, acquiring a minimum enclosing rectangle of a rotary marking box of each target; 2, generating a training set; 3, constructing a deep full convolutional neural network; 4, training the deep full convolutional neural network; 5, detecting a rotating target in the image; 6, post-processing the box of the rotating target; and 7, drawing the final rotation detection boxes of all targets at corresponding positions in the image to obtain a detection result graph. According to the invention, targets with different scales in the image can be better distinguished, false detection is reduced, and the precision of the target box after image target detection is improved.

Description

technical field [0001] The invention belongs to the technical field of image processing, and further relates to a multi-level fusion and corner offset image rotation target detection method in remote sensing image and natural image target detection. The invention can be used to detect rotating targets in remote sensing images and natural images. Background technique [0002] Compared with natural images, remote sensing images have a larger scale variation range of objects in remote sensing images. At different resolutions, the size of the same object varies greatly, and at the same resolution, the size difference of different objects is also large. In order to be able to more accurately distinguish the characteristics of targets of different scales, the reasonable fusion of features at different levels can make each level take into account the target features of other levels on the basis of retaining the target features of the current level, so as to distinguish different s...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to address the defects in the above-mentioned prior art, and propose an image rotation target detection method based on multi-level fusion and corner offset, which is used to solve the problems of low detection accuracy and operating efficiency of the detection process in the prior art The problem of low and large hardware resources

Images