High-resolution remote sensing image airplane detecting method based on high-level feature extraction of depth boltzmann machine

Abstract

The invention relates to a high-resolution remote sensing image airplane detecting method based on high-level feature extraction of a depth boltzmann machine. The method comprises the steps that at first, a picture is divided into a plurality of segments, then scale-invariant feature transformation (SIFT) is utilized for extracting key points in the segments, the key points serve as low-level features of the segments, then a local restriction linear coding algorithm is utilized for coding the low-level features to obtain medium-level features, then the three-layer depth boltzmann machine is utilized for obtaining high-level features of the segments from the medium-level features, then the high-level features are utilized for training a support vector machine classifier, finally the classifier is used for detecting an airplane of the detected picture, and the airplane detection result high in accuracy and robustness can be obtained.

Description

technical field [0001] The invention belongs to the field of computer vision algorithm research, and relates to a high-resolution remote sensing image aircraft detection method based on deep Boltzmann machine extraction of high-level features, which can accurately and robustly detect the aircraft in the image in the remote sensing image database. Background technique [0002] The rapid development of remote sensing technology has enabled many satellite and aerial sensors to provide optical images with high spatial resolution, which have a wide range of applications, such as disaster management, land planning, surveillance and traffic planning. Among these applications, automatic detection of natural or man-made objects is a fundamental task and has attracted increasing research interest. The rich spatial information and detailed structural information contained in optical remote sensing images of aircraft provide us with new opportunities to solve this challenging task. [...

AI Technical Summary

Problems solved by technology

[0003] Feature representation plays an important role in the aircraft detection task of remote sensing images. However, the existing feature representations do not perform well in remote sensing image analysis. The main problem is that it is not enough to extract features based on pixel-based spectral information. Because this ignores the texture space information of the image, it cannot really dig out the deeper structural information of the aircraft in the image. With the advancement of remote sensing technology, high-resolution remote sensing satellites and aerial cameras make it possible to obtain image space and structural information. , many methods begin to use low-level features (such as SIFT, HOG, GLCM) or middle-level features (such as BOW and PLSA) to represent blocks. Although these methods can improve the classification effect and detection accuracy to a certain extent, they will also Faced with some problems, especially the spatial geometric information of the picture that these low-level features can represent is limited and cannot be directly used to describe the structural information of the block. The middle-level features extract the structural information of the spatial region based on the statistical information of the low-level features, but Does not have strong expressive ability in complex background situations

Images