基于嵌入式平台实现的遥感图像城区检测算法

提出了一种城区区域实时检测算法及基于嵌入式平台的实现方式.根据城区是由人造建筑的离散特征聚合而成的特点,城区检测算法划分为原始图像预处理、城区特征点的提取与筛选和基于特征点区域化的城区提取3个环节.通过分析所提出原型算法中各关键环节的处理时间及运算特点,将其分解映射在FPGA+DSP并行异构的嵌入式板卡上,通过流水并行的方式实现.本文提出的城区检测算法已移植于嵌入式板卡在某机载平台上进行了成功的搭载实验,满足实时的城区自动检测需求. 

Urban Area Detection Algorithm of Remote Sensing Image Based on Embedded Platform

A novel urban area real-timedetection algorithm and its realization based on embedded platform were proposed in this paper. According to the characteristic of urban areas which is made up of artificial construction distinguishing feature, the detection algorithm includes three parts. These are image preprocessing, feature points selection and urban area extraction. The algorithm was decomposed and mapped on the embedded platform based on the analysis of its processing time and operation characteristics. And the algorithm was implemented through the way of parallel flowing water.The urban area detection algorithm has been successfully experimented on the airborne platform and can meet the need of real-time and automatically detection.