A method and system for selecting sample length of ship radiation noise based on multiple features

Abstract

The present invention provides a method and system for selecting the sample length of ship radiation noise based on multiple features, including segmenting the ship radiation noise with different sample lengths to form different ship radiation noise sample sets; extracting features from each of the ship radiation noise sample sets, wherein the feature extraction includes extracting multiple features respectively; for each of the ship radiation noise sample sets, each extracted feature is input into a classifier for training, and the recognition rate corresponding to each feature is calculated; for each feature, the associated information of the sample length and the recognition rate is determined according to the corresponding recognition rate, and the associated information obtained for each feature is comprehensively judged, thereby selecting the required sample length. Compared with the traditional method of randomly selecting the sample length, the present invention increases the amount of information carried by the sample and reduces the processing time of the subsequent process.

Images