小目标运动物体检测算法-Pipeline

这是博主的本科毕业论文，实现一个小目标运动物体检测算法-Pipeline，当年获得了优秀毕业论文称号，号称这是哈工大的一个博士课题和项目，现在拿出来分享给大家，感性趣的小伙伴可以加我微信获取代码。

中文摘要

数字图像处理是从20世纪60年代以来随着计算机技术发展而产生、发展和不断成熟起来的一个新兴技术领域，它在理论上和实践应用上都取得了巨大的成就，并引起各方面认识的广泛重视。其中，图像处理在军事、遥感、工业图像处理等大型应用中有不断增长的需求。
“弱小目标”这个概念最初来源于军事领域，在进行空间导弹拦截时，目标与拦截器距离较远（一般几千公里），而在成像平面呈现为一个或者几个像素的大小，称为“小目标”，甚至“点目标”。由于目标太小，很容易被噪声和各种杂波（云杂波、海杂波）淹没，所以又称为“弱目标”。作为航天支撑技术，开展弱小目标理论技术研究已经迫在眉睫。而能够实时有效地进行具有低信噪比和低信杂比的弱小目标的检测、识别与跟踪无疑对于巩固国防、增强国家军事实力、甚至于科学研究都具有深远的意义。
本篇论文给出了一个新的管道检测算法。这个算法可以从具有高噪声图像序列中检测和跟踪单象素运动目标。它利用了运动目标轨迹的平滑性、时间连续性等特点，在序列图像中，沿着观测窗口进行目标能量的累加。然后通过计算形心得到目标的位置信息。管道方法突破了其他可以完成相似任务的方法只能探测直线轨道的限制。并且算法是完全并行分布处理，因此，具有较高的检测率。能够满足实时性的要求，能够检测信噪比（snr<=2）的弱目标，是高噪音环境下检测和跟踪具有任意轨迹的运动小目标的理想方法。
这次毕业设计在当前最流行的开发环境vc++ 6.0下实现了目标检测算法—Pipeline算法，并且以友好的图形界面演示了Pipeline 算法的全过程，并提供了全部相关的中间结果，以供查看。

Abstract

At the end of 20th century, with the development of computer science, image manipulation begins to occur, develop, and become a new technology. It obtains a great success theoretically and practically. Therefore, it achieves abroad recognition. Thereinto, image manipulation is needed in military affairs, remote sensing, industry image manipulation and so on.

The concept " weak and small target" roots in the military realm. When guided missile is heading off, the target is far away from interceptor (about several meters), therefore, the target is mapped onto "small target" or "one pixel". The target is so small that it is easy to be submerged in noisy and clutter, therefore, we call it "weak target". As the main technology of spaceflight, studying the theory of "weak target" is very important. Detecting and tracking "weak and small target" with lower noise levels is very important in many fields such as national defense, military affairs and science.

This paper presents a new pipeline method for detection and tracking of pixel-sized moving targets with unknown trajectories from a time sequence of highly noisy images. The Pipeline Target Detection Algorithm (PTDA) uses the temporal continuity of the smooth trajectories of moving targets, and successfully detects and simultaneously tracks all the target trajectories by mapping them from the image sequence onto a single target frame. The pipeline method breaks the constraint of straightline trajectory that most other algorithms require for similar tasks. The algorithm is a complete parallel distributed processing (PDP) type process, and therefore is highly time efficient—ideal for real time detection and tracking of arbitrary target trajectories in high noise environments.

This program is achieved under vc++ 6.0, which is most popular programming language. It is a demo in graph interface containing Pipeline Algorithm. In addition, it supports all middle results in order to show the algorithm to user clearly.