中文摘要：

针对在三维表面上均匀采集的少量数据点,提出一种基于压缩感知与最小二乘支持向量机(LSSVM)的三维组织表面重建方法.通过结合采用拟合与插值方法得到与待重构表面数据相同数目的数据点集,采用离散余弦变换(DCT)分别得到其三维坐标的稀疏系数,用设计的自适应观测矩阵进行观测,并选用正交匹配追踪算法作为重构算法,最后采用LS-SVM回归预测模型对压缩感知重构结果进行修正.实验结果表明:该重建方法得到的组织表面数据误差小,能保持在1mm左右,重建表面光滑,为基于虚拟现实的虚拟手术系统提供了精确的表面数据模型.

英文摘要：

A method of 3D tissue surface reconstruction based on compressed sensing (CS ) and least squares support vector machine (LS -SVM ) was proposed for a small amount of uniformly sampling data points on 3D surface .Firstly ,the same amount of data points with the surface to be reconstruc-ted was obtained by using fitting and interpolation method . Then , the discrete cosine transform (DCT ) was adopted for the 3D coordinate sparse representation respectively ,and the designed adap-tive observation matrix for signal observation .The orthogonal matching pursuit (OMP) was used as reconstruction algorithm .Finally ,the results of compressed sensing (CS) reconstruction were correc-ted by LS -SVM regression prediction model .Experimental results show that the tissue surface recon-struction data error based on the method proposed is small ,and the reconstructed surface is smooth , w hich can provide accurate surface data model for virtual surgery system based on virtual reality .