Target detection method and system based on high-resolution optical satellite remote sensing images

Abstract

The invention relates to a target detection method and system based on high-resolution optical satellite remote sensing images. The method includes obtaining marked target positive samples and background negative samples to form training samples; extracting a plurality of different weak feature channels for the training samples, according to Multiple different weak feature channels acquire candidate areas; acquire the context scene of the candidate area, extract features from the candidate area and the context scene of the candidate area, and fuse the extracted features to form candidate area features; train the training samples, Obtain a classifier; use the classifier to classify the features of the candidate area to obtain the target area containing the target; perform deduplication processing on the target area to obtain the detection target. The invention realizes the target detection on the remote sensing image with enlarged format, and optimizes the target detection effect of the target with very close distance and the target with unusual aspect ratio.

Description

technical field [0001] The invention relates to a target detection method, more specifically to a target detection method and system based on high-resolution optical satellite remote sensing images. Background technique [0002] Target detection is to solve whether there is a specific target in the image and where the target is. It is the basis of image retrieval, image interpretation, target recognition and tracking and other fields. It has been a research hotspot in academia and industry in recent years. Many specific target detection systems based on natural images or videos have been put into industrial use, such as pedestrian flow statistics based on pedestrian detection, face detection beauty cameras, and automatic driving technology for obstacle detection such as vehicles and pedestrians. [0003] There are currently two commonly used methods for target detection, one is target detection based on natural images, and the other is target detection based on remote sensin...

AI Technical Summary

Problems solved by technology

However, there are usually many targets in high-resolution remote sensing images, which may be scattered in various parts of the image, and the targets to be detected are relatively small and close to the background. After scaling the image, the small-scale targets lose many details, and due to the The background is relatively close, which also increases the difficulty of detection. Therefore, the existing target detection methods on natural images cannot perform target detection on extended large-format remote sensing images.

Images