针对大视场、小目标、多目标特点下基于数字图像的飞行目标姿态判读任务，选用顶帽运算抑制背景.提取目标骨骼，利用改进Radon坐标系计算极限角分辨率.通过在传统Hough变换后加上一次改进局部Radon变换的方法提取多目标姿态参量，解决了线段位置检测问题,同时消除了小目标、多目标情况下，单纯使用Hough变换处理时产生的误判现象.提出一种变步长快速游标尺度逼近算法,在目标尺度提取上获取了较高的判读准确度.

The background suppression cap calculations is chosen based on digital image interpretation of goal-attitude mission in the context of large field,small goals,multi-targets.The aim bone is extracted,and the limits of angular resolution are calculated by improving Radon coordinate system.The parameters of multi-targets attitude can be extracted by transforming the traditional Hough plus improvement of a local Radon transform,thus resolve the problem position detection segment.At the same time,it could eliminate the miscarriage of justice caused by simply using the Hough transform in the context of small goals and multi-targets.A variable step-size cursor Express-scale approximation algorithms is put forward in order to get a higher extraction of interpretation accuracy in the extraction of aim-scale