动态图像的拼接与运动目标检测方法的研究

动态图像运动目标检测是图像处理中的热点,但动态图像的识别范围却成了目标检测的限制,针对此问题,本文提出了一种利用图像拼接技术扩展图像识别范围、并在此基础上完成运动目标检测的方法。在图像拼接中采用了SURF图像匹配算法,运动目标识别利用背景差分法,实验中使用的是开源的Linux操作系统、以及为图像处理提供了大量算法和函数的Open CV软件开发库。针对不同分辨率、不同角度采集的图像进行了实验研究,结果表明,可以在较好满足图像识别范围的同时,明确地检测出运动目标的相关信息。同时,本文提出一种通过图像拼接实现扩展运动目标检测的方法,满足了实时性要求,达到了增加图像清晰度的目的,但是,在摄像设备与场景之间的相对运动方面还存在着有待解决的问题,这将成为今后研究的重点方向。

The Research of Dynamic Image Mosaic and Moving Object Detection Method

The moving object detection of dynamic image, in the modern world is a key point, a difficult point and also a hotspot in image processing, it has a lot of useful application, such as keep a lookout over traffic situation in intelligent transportation, the precise guidance in military affairs and so on, but the recognition range of dynamic image has been the limitation of moving object detection. In order to solve this problem, this paper have made a research and put forward a kind of new method that using image mosaic technology to expand the scope of image recognition, and then based on the wide angle image we have get to finishing the detection of moving object. In this experiment, first we considered the problem of real-time and then we choose the SURF（speeded up robust transform） as an image matching algorithm instead of the SIFT（scale invariant feature transform） which is usually adopted in image mosaicking, because of the SURF algorithm can provide enough characteristic points and on other hand it＇s working speed is faster than SIFT algorithm, and after the image mosaic we start to do the detection work about the moving object, at this time we took advantage of the background difference method, which can help us detect the moving object completely and quickly. Second, we considered the problem of practicability and then we choose the open source systemLinux as the operating system and the software library which named Open CV is used in this experiment as the tool of image processing which can provided a large number of algorithms and functions, the reason that we choose this software is that all of them are free. We can get them through the internet easily. We use the image that we take about one sight from different angle, different distance and the image with different resolution rate to do this experiment, so that we can have a general result. At the end of this experiment, the study results show that, we can get the dynamic image that satisfy the requirement of real-time and the enhancement o