A moving target detection method for remote sensing image sequences based on hierarchical saliency analysis

Abstract

The present invention provides a method for detecting moving targets in remote sensing image sequences based on hierarchical saliency analysis. The specific process is as follows: firstly, a dual-mode single-Gaussian model based on motion compensation is used to extract time-domain saliency information on a global scale to obtain Time-domain saliency map; roughly extract the motion region from the time-domain saliency map as a candidate region; secondly, on the basis of the candidate region, carry out spatial-domain saliency analysis at the pixel level and super-pixel level respectively, and obtain the spatial-domain saliency Figure; finally, the obtained global temporal saliency map and local spatial saliency map are linearly fused to obtain the spatiotemporal saliency map, and the moving object detection result is obtained by threshold segmentation. This method can quickly locate the moving target from the coarse scale to the precise description at the fine scale, which can effectively improve the computational efficiency of the moving target detection algorithm.

Description

technical field [0001] The invention belongs to the technical field of remote sensing image processing, and in particular relates to a method for detecting moving targets in remote sensing image sequences based on hierarchical saliency analysis. Background technique [0002] Automatic detection of moving objects in remote sensing image sequences is one of the more important research directions in the field of remote sensing. In recent years, with the development of UAV technology, UAVs have been increasingly used in the field of remote sensing, such as geological disaster investigation, agricultural crop detection, etc. Compared with satellite remote sensing technology, UAV remote sensing technology has the advantages of low cost and high timeliness. It can carry a variety of sensors, continuously observe areas of interest, and obtain a large amount of dynamic remote sensing data. Quickly and accurately extracting the information of moving targets from these remote sensing ...

AI Technical Summary

Problems solved by technology

However, spatio-temporal saliency models usually have a contradiction between model complexity and final detection results: complex models have better detection results, but time-consuming calculations; simple models can meet real-time requirements, but poor detection results

[0007] In the moving target detection method for the moving observation platform, the optical flow method and the method based on motion compensation usually cannot meet certain real-time requirements, and the information of the target space domain is insufficiently utilized, resulting in low accuracy of the final detection result; The method based on spatiotemporal saliency is a new idea of ​​moving target detection, which can comprehensively utilize spatiotemporal saliency to improve the detection effect, but there is usually a contradiction between model complexity and detection effect

Images