景像匹配仿真的一种新方法

分析了影响景像相关匹配精度的各种误差因素，并给出它们的误差仿真模型，该模型可以适应飞行姿态、飞行器特性，天候等条件变化时的匹配概率估计，并可以与图像特征参数分析方法相结合。实验验证了该模型的有效性，且可以快速地给出影像图的适配性估计图，用于飞行器的航迹规划。

A New Algorithm for DSMAC Simulation

Digital scene matching area correlation (DSMAC) is a technique that has been proven to be highly successful as a navigational aid for autonomous,unmanned guided vehicles.In order to evaluate the scene correlation suitability or a performance of DSMAC algorithm,it is necessary to make a scene simulation model. 　　Terrain contour matching (TERCOM) provides real-time fight control data and one dimensional height data for period near real-time position update of the vehicle's inertial navigation system (INS).Because the error of TERCOM is not sufficient to decide the area,the more exact two-dimensional navigation update technique DSMAC is used［1］. 　　It is well understood that the periodic update and the quantity of DSMAC navigation updates required is the function of:① the performance of gyro in the INS.② the flight time of mission,and ③ the precision with which the unmanned vehicle must overfly a reconnaissance point or target location.However,less understood are the processes by which the scene correlation effectiveness of a candidate point can be predicted and ranked using a multi-parameters figure-of-merit (FOM) and how the selection of an individual FOM and the resulting accumulated series of scene suitability FOMS can be related to all mission reliability. 　　Computer models currently available to predict DSMAC scene suitability use a single parameter ranking system,such as signal-to-noise (S/N) ratio.Current models also assume the aircraft flies straight and level,and the optimum heading,while over the scene reference area.These limitations require the introduction of an adaptive flight path modification capability in future robotic aircraft,which provides the ability for defense avoidance,or inflight re-routing.This capability will certainly require a more robust DSMAC scene selection process that is able to accommodate variations in aircraft altitude,attitude,orientation and initial position uncertainty relative to the potential scene.Current model is not enough to evaluate a real vehicle. 　　The subject of this paper is the development of a Monto-Carlo simulation technique capable of modeling a robust DSMAC selection process,as discussed above,which characterizes scene suitability using a multi-parameter FOM .The Monte-Carlo simulation described in this paper is modular and expandable to support future studies involving other altitude reference baseline (e.g.global positioning system),other height finding instrument (e.g.LASER altimeters),and other map reference derivatives (e.g.interferometric SAR maps). 　　In this paper,we propose a DSMAC simulation model.A Monte-Carlo simulation is produced to evaluate the potential DSMAC success rate under various scenarios and correlation algorithm.The simulation details are provided together with some preliminary result to demonstrate the feasibility of the proposed technique.It has been shown that the proposed technique can provide a cost-effective enhancement to the TAN-based mission planning process.For the future work,it can be transplanted to parallel process.It can improve the speed of calculation evidently.