移动目标检测与跟踪技术的研究及matlab实现.pdf

移动目标检测与跟踪技术的研究及 Matlab 实现

摘要

在现实生活中，人通过眼睛能够看到生动的外界的光信息，人眼对移动的信息也格

外的敏感，因为移动的信息包含的可能更多的有效信息。可是好多现实场景人是无法到

达的，这时就需要机器视觉来辅助人了解那些人没法到达的地方的实际情况，如高空中

的移动目标监测，海底目标监测等各种实际需求。因此研究机器视觉显得格外重要。移

动目标检测与跟踪作为机器视觉中的重要分支，研究意义十分重大。

仿真实现，简单介绍了数字图像处理的基本概念及相关知识点，着重分析移动目标检测

和跟踪的方法，运用一种改进的帧间差分法-三帧差分法实现移动目标检测，运用移动目

标跟踪中的两种算法的结合算法实现跟踪。实验证明所用方法均能提高了检测和跟踪的

效果。简要内容如下：

1. 为了对后期移动目标检测及跟踪更加方便的理解，本文对数字图像处理相关的概

念及视频的采集过程进行较为详细的了解，介绍了数字图像处理的相关概念，数字图像

处理中常见的处理方法（灰度变换、均值滤波、中值滤波、二值化）以及运用 matlab 进

行图像的采集及视频的保存。

2. 在本文中移动目标检测服务于下一步移动目标跟踪，为了提高跟踪效果，需要提

高移动目标检测的效果，本文对移动目标检测常用的三种算法进行了详细的介绍，并进

差，本文运用 均值漂移算法与卡尔曼滤波算法结合的方法进行移动目标的跟

踪，可以有效的解决目标被遮挡的情况。

关键词：视频采集，移动目标检测，移动目标跟踪，均值漂移，卡尔曼滤波

So the study of machine vision is particularly important. Moving target detection and tracking

The moving target recognition and tracking is for the video information stored in the

computer, and the video information can also be regarded as constituting the image sequence

on the time axis by the frame image. This paper moves the target detection and tracking in the

fixed camera to capture the video in the case of the case. Mainly including mobile target

detection and tracking, detection is the video whether there is a moving target for intuitive

detection, tracking is based on the detection of the target to determine the target in the adjacent

This paper mainly introduces the basic concepts and related knowledge points of digital

1. In order to understand the characteristics of the digital image processing and the

acquisition process of the video, the related concepts of digital image processing and the

common processing methods in digital image processing are introduced in order to understand

the post-moving target detection and tracking more easily. Gray-scale transformation, mean

filtering, median filtering, binarization) and the use of matlab for image acquisition and video

and the mean shift algorithm of the moving target tracking are introduced by using the detected

target and complex background, this paper uses the Mean Shift mean shift algorithm and the

Kalman filter algorithm to track the moving target, which can effectively solve the problem that

Key words: video capture, moving target detection, moving target tracking, mean drift, Kalman

filter