mytracking.m

基于特征目标跟踪程序 matlab 程序

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% PROJECT:
%     Fast Feature Tracking 
% DATE/Version:
%     2008.4- V 1.0
% Description:  
%     The provided p code constitues the Fast Feature Tracking algorithm as
%     implemented in Matlab. Run the MyTacking.m filte to see the algorithm
%     work on the given videos.
%     NOTE: This is by no means an optimal implementaion, there are many
%           redundent calculations that can be eliminated.It can run over 25
%           Frames per s after optimization. The Tracking Result is in
%           current directory!
%           If you the errors happen during Tracking,please type "clear
%           all" in the Command Window,then you can see the Tracking Result in
%           current directory!
% References:
%     Now the optimal C code has ran better results not only in Windows but also in DSP Platform.
% Project file list: 
%     MyTracking.m
%     ObjectTracking.p
%     Location.p
%     FeatureSelect.p
%     FeatrueFind.p
%     FeatureCheck.p
%Contact:[e-mail] QQ:418123705
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%需要注意的是算法是跟踪序列图象，需要将视频转换为单帧的序列图象，且命名是按照001.jpg表示第一帧图象，
%500.jpg表示第500帧图象，且这里已被限制最多跟踪500帧图象，图象格式为.jpg格式，目标大小也被限制为不
%大于60*60，不小于8*8，跟踪效果与是否准确初始化目标大小以及背景复杂度有一定联系，能够接受目标发生平
%移旋转以及仿射变化，同时对于目标较小尺度变化也能够接受，同时对图象在低信噪比也能够实现稳定的跟踪，
%对抖动不敏感，相对与许多传统的跟踪算法而言，本算法更能适应跟踪的一般性。%算法运算量并不大，能够在
%WINDOWS上实现目标的稳定跟踪，对于直接移植到DSP上不需要优化代码便能够达到实时，具有较好的跟踪结果。

close all;clear all; clc     %清空当前内存变量，关闭图象窗口，清空命令行
cd('d:\MATLAB08\work\Test2') %将工作目录更改为当前目录
Hight=30;                    %目标垂直高度设置
Width=20;                    %目标水平宽度设置
ObjectTracking(Hight,Width); %跟踪开始