一种基于行程编码的图像背景重构

背景差法是一种重要而且实用运动目标检测方法。该算法的难点在于如何进行背景图像的重构。本文针对该问题,提出一种基于行程编码的背景图像重构算法,即在假设背景图像以最大概率出现在图像序列中的前提下,选择频率最高的图像作为背景图像进行背景重构。利用序列图像时间上的相关性进行行程编码,用编码数据进行背景图像点得重构,使重构具有实时性和节省运算空间的特点。在慢运动目标的图像序列中,即观测时长较长时,利用本文算法进行实时背景重构具有明显的优点。仿真结果表明,该算法能够准确地重构背景,从而实现对运动目标的完整提取,以便进一步识别或跟踪。

An Image Background Reconstruction Based on Run Length Encoding

The background subtraction is an important and practical target detection method.The problem is how to perform the background image reconstruction.This paper presents a method based on run length encoding background image reconstruction algorithm,namely,under the assumption that the background image appears in an image sequence to the maximum probability,choose the highest frequency of the image as the background image background to reconstruct the background.Using sequential images of temporal correlation coding itinerary and using coded data background image point to perform reconstruction,the reconstruction has the advantages of real time and saving operation space characteristics.In the image sequences in slow moving target i.e.when the observing time is longer,the real-time background reconstruction algorithm in this paper has obvious advantages.The simulation results show that the algorithm can accurately reconstruct background and completely extract the moving goals to perform further recognition and tracking.