Remote sensing image weak and small target fusion multi-level feature target detection method

Abstract

The invention relates to a remote sensing image weak and small target fusion multi-level feature target detection method, and the method improves the detection precision of a remote sensing image weak and small target through an end-to-end target detection network based on a convolutional neural network and the multi-level features of the remote sensing image weak and small target. The invention provides a detection method for weak and small targets based on the characteristics of small size, such as dozens of pixels and weak characteristics of the weak and small targets of remote sensing images in actual engineering. According to the method, target features are extracted layer by layer through a convolutional neural network, cross-level channel fusion target features, position attention mechanism aggregation target features and double-branch feature map prediction targets are performed, and an end-to-end target detection network is formed, so that the purpose of performing high-precision and high-speed detection on weak and small targets in remote sensing images is achieved. Experiments show that the recall ratio and precision ratio of remote sensing image weak and small target detection can be effectively improved.

Description

technical field [0001] The invention belongs to the technical field of computer vision and machine learning, and in particular relates to a fusion multi-level feature target detection method for weak and small targets in remote sensing images. Background technique [0002] With the development of remote sensing image target detection field, faint target detection has gradually become the research focus of remote sensing earth observation. Weak and small targets show the characteristics of weak features and small size in remote sensing images. The weak features are mainly reflected in the unclear outline of weak and small targets, the lack of prominent texture features, and high similarity with adjacent background features. The small size is mainly reflected in the weak and small targets in remote sensing images. The number of pixels in the remote sensing image is small, such as dozens of pixels, so only a small number of effective features of weak and small targets can be ex...

AI Technical Summary

Problems solved by technology

[0005] In order to overcome the problem of low detection accuracy in the detection of weak and small targets in remote sensing images by conventional target detection methods, the present invention proposes a fusion multi-level feature target detection method for weak and small targets in remote sensing images

Images