基于空间相关性的图像缝合算法

Image Stitching Algorithm Based on Spatial Correlation

作为一种新型的区域观测技术,无线多媒体传感器网络(WMSN)获得了国内外研究者的高度关注。在密集部署的无线多媒体传感器网络中,不同的相机节点所观察到的视觉信息存在较大的冗余和空间相关性。为了适应无线多媒体传感器网络的带宽有限、计算能力有限、功耗有限的缺点,利用WMSN节点的空间几何关系以及相机成像原理,获得图像之间的相关系数,根据信息熵的原理建立相机节点的选择机制。在此基础上,利用相关系数对选择的相机采集的图像进行缝合。降低了复杂度和计算量,使之更适合WMSN节点的应用。

Line Multimedia Sensor Networks （WMSN） wins high attention from researchers both home and abroad as a new type of regional observation technology. In the dense deployment wireless multimedia sensor network, the visual information different camera node observes is redundant and has spatial correlation. In order to adapt to the disadvantage of WMSN such as limited bandwidth, calculation ability and power dissipation, this paper uses WMSN node＇ s space geometrical relation- ship and camera imaging principle to acquire correlation coefficient between the images, and establi- shes a camera node selection mechanism according to the theory of information entropy. Based on this, this paper uses correlation coefficient to stitch the images collected by cameras that has been chosen before. This method greatly reduces the complexity and computation, and makes it more suit- able for the application of WMSN nodes.