A Novel Object Detection Approach for Satellite Imagery Based on Danger Theory

Abstract

This paper discusses the relationship between Danger Theory and image object detection, and interprets the produce of the image object detection using Danger Theory. In the paper, the image object detection is mimiced as the produce that biology immune system detects danger antigens, where the interested object are regarded as ldquodanger antigensrdquo and object detectors are regarded as antigen present cells. The architecture of Danger Perception Network (DPNET) and learning algorithms are described in the paper, and the approach of applying Danger Theory in image target detection is also presented. Experiments of vehicle and airplane detection in high resolution satellite images are given to illustrate the feasibility and effectively of proposed method.