摘要
由于视觉计算理论的局限、硬件水平的限制 ,研究全自动的遥感图象解译方法还不现实 ,因此开发了一种计算机辅助图象判读系统。在图象中感兴趣的区域指定一个核 ,利用有约束的区域增长法来分割目标 ,并用数学形态学的结论对分割出的目标进行处理 ,获得目标的分割图象。该系统能帮助判读员在复杂的航空图片中迅速地勾画出某一目标的轮廓 ,确定目标的种类。进一步的工作还可以让计算机完成目标的识别 ,把可能的识别结果按概率的大小顺序显示在屏幕上 。
Aiming at processing remote sensing images quickly and accurately, we present a new interactive system for image segmentation by using an improved region growing algorithm and morphology. Section 1 deals with images of even intensity. We first gave two criteria (eqs.1 and 2) for determining if a point belongs to a particular region. Based on the criteria, we put forward an algorithm, we call it algorithm I(steps 1 through 10), for the segmentation of such images. Figs. 1 and 2 give the simulation results of algorithm I. Section 4 deals with images of uneven intensity by revising criterion 1 in section 1. The revised criterion is given by eq.(3). Making some adjustment in step 5 in algorithm Ⅰ, we get algorithm Ⅱ. Fig.3 gives the simulation results of algorithm Ⅱ. Our system has met with preliminary success in engineering applications.
出处
《西北工业大学学报》
EI
CAS
CSCD
北大核心
2000年第2期289-292,共4页
Journal of Northwestern Polytechnical University
基金
国家自然科学基金资助! ( 69772 0 3 0 )
