摘要：For the automatic recognition of red tide algaeimages in the marine environment,current predic-tion methods fail to consider the local minimumproblem in nonlinear classification research,whichleads to the long time-consuming,low accuracy andhigh false alarm rate of the automatic recognition ofred tide algae images.In order to effectively solvethe above problems,in this paper,we propose an au-tomatic recognition method of red tide algae in themarine environment based on feature selection.Thismethod uses image processing technology to extractthe morphological and texture features of red tide al-gae images in the marine environment.And,it obtains the best feature combination through image ac-quisition,image preprocessing,segmentation,fea-ture extraction and feature selection.The artificialneural network is used to construct a classifier toclassify the red tide algae image,which can effec-tively realize the automatic recognition of the redtide algae image.The simulation experiment resultsshow that the proposed method can effectively im-prove the recognition accuracy and reduce the falsealarm rate.
