论发展遥感图像应用智能处理技术

任何对地观测系统获得的原始图像数据实质上都是地球表面电磁波谱特征的综合反映。经地面站系统处 理得到的常规图像产品仍存在大气、地形等多种随机因素导致的几何畸变与辐射失真 , 并且多种地物目标信息混 杂 , 甚至相互迭掩 , 彼此抑制。所以 , 严格地说 , 这些图像只是一种信息源 , 属“数据”范畴 , 必须对其进行再 处理才能得到适应不同专业需要的优质图像和有用信息。至今 , 从事遥感图像处理的科技人员大多出身于电子、计 算机等专业 , 故对于不同地理景观、不同季节、不同传感器所获得的图像数据 (一串灰度数码) , 往往很难体会到 其中包含着的极为丰富而复杂的生物、地学内容。一个能出色完成一般图像处理任务的杰出图像处理专家 , 不一 定能胜任各种不同应用目的所要求的图像处理任务。常常使得海量遥感数据难以得到有效的利用。据此 , 笔者认 为应把遥感图像处理分解为系统处理与应用处理两大类 , 它们是既相互衔接又彼此有别的技术分支。前者只从图 像总体要求出发 , 着眼于改善、提高图像整体质量 ; 后者则必须密切关注具体的应用目的与对象 , 形成融入了地 学分析思路、并精确反映遥感成像机理的图像应用处理技术。文中给出了发展遥感图像应用处理技术的流程框图 , 列举了森林火灾、虫灾监测 , 洪涝灾情快速反应 , 城市扩展 , 低产土壤清查 , 土地利用动态监测等多项任务中的 成功实践。

On the Intelligent Processing Techniques of Remote Sensing Application

Raw images obtained by whatever earth observation systems are the comprehensive reflection of magnetic spectrum characteristics of surface objects The primary images produced through systematic processing in remote sensing ground station still contain various errors in geometry and radiation because of the influence of atmospheric and topographic conditions And the information from various objects mixes together and influences each other Therefore, what the primary image provides is only data instead of information This data must be reprocessed to get high quality image and useful information suitable to different special applications However, the specialties of most researchers who are engaged in remote sensing image processing are computer, electronics etc They have not enough knowledge in application fields and can’t understand well the rich and valuable information which the remote sensing data of different sensors and different dates contains Apparently, even an excellent expert on image processing can’t complete processing tasks effectively to meet the demands of various applications Thus, up to now, the huge and valuable remote sensing data haven’t been used fully and effectively We think that the remote sensing image processing should be divided into two parts, systematic processing and application processing They are two parts of remote sensing processing, being both closely related and differing from each other The purpose of systematic processing is to improve the quality of image on the whole While that of application processing is, according to concrete application purpose and task, to carry out the comprehensive and object oriented processing so as to analyze and extract thematic information accurately and effectively This paper discussed the general method and procedure of remote sensing image application processing, and showed some successful examples of using it in forest fire detecting, rapid response of flood, city expansion investigation, soil resource survey and management, and dynamic monitoring of landuse.