Parallel Processing Method for Feature Extraction of Remote Sensing Impacts

The main problems faced by remote sensing image processing are the large amount of data to be processed and the complexity of imaging scenes. On the one hand, regardless of the type of remote sensing image, it faces the problem of massive data computation during processing, while traditional remote sensing image phenomenon processing mainly relies on the prior knowledge of domain experts and has a low degree of automation. On the other hand, remote sensing images have a large coverage area and contain a large amount of feature information, with multiple gray levels and a large amount of information. Compared to ordinary images, feature extraction from heterogeneous images is more difficult. Therefore, achieving heterogeneous image information matching is a daunting task. This article presents a parallel heterogeneous image feature extraction system. This system can support grid window based feature extraction and analysis of optical and Synthetic Aperture Radar (SAR) images, using multi CPU parallelization. The extracted features are evaluated and sorted using a decision tree model, and the evaluation results are verified to be close to manual annotation through semi physical simulation. Its accuracy is greater than $80 \%$, and the computational efficiency is more than 4 times that of single CPU processing.