A ground-based cloud classification method based on multi-cues and multi-modal fusion deep network

Abstract

The embodiment of the present invention discloses a ground-based cloud classification method based on a multi-cue multi-modal fusion deep network. The method includes: preprocessing an input ground-based cloud sample to obtain a multi-cue multi-modal fusion deep network input; Transfer to the multi-cue multi-modal fusion deep network training model, and train to obtain a multi-cue multi-modal fusion deep network; extract the global visual features, local visual features and multi-modal features of each input ground-based cloud sample, and fuse to obtain each Input the final feature representation of the ground-based cloud sample; train the support vector machine classifier to obtain the ground-based cloud classification model; obtain the final feature representation of the test input ground-based cloud sample, and input it into the ground-based cloud classification model to obtain the classification result. The invention has the capability of fusing heterogeneous features, can effectively learn visual information and multi-modal information, extract higher discriminative global visual features, local visual features and multi-modal features, and improve the accuracy of ground-based cloud classification.

Description

technical field [0001] The invention belongs to the technical fields of pattern classification, meteorological science and artificial intelligence, and in particular relates to a ground-based cloud classification method based on multi-cue multi-modal fusion deep network. Background technique [0002] Ground-based cloud taxonomy has been extensively explored by researchers over the past few decades. Ground-based cloud classification is important for many practical applications, including climate prediction, air traffic control, weather monitoring, etc. Today, ground-based cloud classification still mainly relies on professional observers, while manual observation consumes time and manpower, and is affected by observer experience and subjective factors, resulting in ground-based cloud classification results that often vary from person to person. Therefore, it is urgent to propose accurate automatic ground-based cloud classification methods. [0003] Most traditional methods ...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to solve the difficult problem of ground-based cloud classification. For this reason, the present invention provides a ground-based cloud classification method based on multi-cue multi-modal fusion depth network

Images