Low-sample size remote sensing image recognition based on a multihead attention integration network

AbstractThe traditional target recognition method for the remote sensing image is difficult to accurately identify the specified targets from the massive remote sensing image data. Based on the theory of multitemporal recognition, an automatic target recognition method for the remote sensing image is proposed in this article. The proposed recognition method includes four modules: automatic segmentation of multitemporal remote sensing image, automatic target extraction of multitemporal remote sensing image, automatic processing of multitemporal remote sensing image, and automatic recognition of multitemporal remote sensing image. The automatic segmentation of the image target is introduced. The effectiveness of the segmentation technology is verified through the kernel function bandwidth algorithm. Linear feature extraction is used to extract the segmented image. The image extraction processing is described, which includes image profile analysis, image preprocessing, image feature analysis, the region of interest localization, image enhancement processing, recognition processing, and result output. According to the theory of pattern recognition, three different feature recognition images are given, which are partial separable recognition, weakly separable recognition, and fully separable recognition, and then, a new image recognition method is designed. To verify the practical application effect of the recognition method, the proposed method is compared with the traditional recognition method. Experimental results show that the proposed method can accurately identify the specified objects from the massive remote sensing image data and has a high potential for development. This article has an important guiding significance for image recognition.

Download Full-text

Feature Selection for Object Recognition by Property Related Analysis

Key Engineering Materials ◽

10.4028/www.scientific.net/kem.480-481.323 ◽

2011 ◽

Vol 480-481 ◽

pp. 323-328

Author(s):

Yong Mei Zhang ◽

Li Ma

Keyword(s):

Remote Sensing ◽

Feature Selection ◽

Object Recognition ◽

Association Analysis ◽

Image Recognition ◽

Remote Sensing Image ◽

Decision Algorithm ◽

Multi Agent ◽

Recognition Efficiency ◽

Selection For

The realization of object recognition is discussed, the intrinsic relation between factors through high step association analysis is presented to mine the various features using property related analysis, the sensitive recognition features are chosen to improve the recognition efficiency. A multi-agent object recognition model(MAORM) is established, which combines concurrency research results and the specific characteristics of multi-sensor remote sensing image recognition. A majority-decision algorithm based on multi-agent is proposed. Experiment results show the system can effectively identify the bridges, wharfs, ships and so on. Compared with a single remote sensing image, the system can distinguish targets with higher recognition.

Download Full-text