基于缝合技术的图像传感器网络消除冗余数据方案

Redundancy elimination scheme based on stitching technique in image sensor networks

为了减少密集部署的无线图像传感器网络中相机节点的冗余数据,提高网络生存周期,利用图像缝合技术在簇头节点找出相机节点采集图像的重合区域,并告知相机节点,相机节点过滤掉与相邻节点重合的冗余图像数据,簇头节点对接收到的图像进行重建并拼接成全景图.提出了一种面向图像的节点相关系数计算方法,解决了现有相关系数计算方法中要求准确获知节点空间位置的问题.仿真实验结果表明,该方法使相机节点能有效地去除冗余数据,簇头节点能重建出高质量的全景图,而且面向图像的相关系数很好地反映了相机节点的相关性.

For the purpose of reducing the redundant data of camera nodes and prolonging network lifetime in densely deployed image sensor networks,the image stitching technology is used in cluster heads to locate the overlapping area of the images that camera nodes captured,then,the overlapping area information is transmitted to camera nodes.Thus,camera nodes reduce the overlapping image data and cluster heads reconstruct the received images and stitch them.Moreover,a new image-oriented node correlation coefficient computing method is proposed to solve the problem in current method that precise node position information is required.Simulation experiments indicate that the proposed method enables camera nodes to efficiently eliminate redundant image data and cluster heads are able to reconstruct panoramic images of high quality.Moreover,the image-oriented node correlation coefficient computing method describes the correlation of camera nodes preferably.