Target Detection for Remote Sensing Image Based on Multiple ERP Components

Abstract

Target detection technique based on event-related potential (ERP) has become a new method to solve the problems of small target detection with complex background in remote-sensing images. Multiple ERP components reflect different stages of the visual information processing, previous studies usually used features of single ERP component to detecting targets. In this paper, a target detection model was built based on multiple ERP components evoked in rapid serial visual presentation (RSVP) of remote-sensing images. Experimental results showed that combing the spatiotemporal features of multiple ERP components could improve the detection accuracy, where the mean AUC could reach 89%.