中文摘要：

分离出无线电层析成像方法中对称链路中的冗余信息后转换为噪声信息，通过降低重建图像中的噪声来实现图像增强。分析RSSI变化量与阴影衰落的线性模型后，利用对称信息对系统矩阵降低维度，将投影矩阵分解为A组正链路和B组逆链路冗余投影值两部分。由于两组投影值双向差的线性反投影析出了噪声信息Nab和Nba，所以用A组和B组数据的线性组合能抑制原重建图像的噪声达到增强图像的目的。选用JN5139芯片构成Zigbee无线传感网节点，采用16个节点围成23．04m^2正方形实验环境，分别采集了1人和2人站在区域内的测量数据。实验数据成像结果表明，A组和B组数据均能完整地重建图像，两组投影值双向差析出的是噪声信息，当选用α=40，β=5时，A组和B组数据的线性组合重建图像能清楚地确定人的站立位置。该算法不仅能通过对系统矩阵降低维度来加快运算速度，而且还能通过A组和B组数据的线性组合来增强重建图像。

英文摘要：

Separated redundant information of symmetrical link is converted into noise in the radio tomographic imaging; reconstructed image is enhanced by reducing the noise. After linear models of RSSI variation and shadow fading are analyzed, system matrix dimension is reduced for symmetric information and the projection matrix is decomposed into 2 sections：link group A projection value and inverse link redundancy group B projection values. The noise information Nab and Nba are precipitated by linear back ward projection of bidirectional difference of two sets of the projection values; noise is suppressed and reconstructed image is enhanced by linear combination of data group A and group B. Selecting JN5139 chipas Zigbee node in the wireless sensor network, 16 nodes are deployed around 23.04 m2 square experimental environment and then, measurement data are collected when 1 person and 2 persons stand in the area. Experimental data imaging results show that the group A and group B data can be used to reconstruct image completely. Bidirectional difference of two sets of projection values precipitates noise information and the reconstructed image can clearly determine the person＇s standing position by linear combination of data group A and B when alpha equals to 40 and beta equals to 5. The algorithms not only speed up computations through reducing the dimension of the system matrix, but also enhance the reconstructed image by a linear combination of data group A and B.