本文選題：TSVD　切入點：截斷參數(shù)　出處：《中國礦業(yè)大學學報》2017年06期

【摘要】：截斷奇異值法(TSVD)通過截斷參數(shù)截掉病態(tài)矩陣中較小的奇異值來改善模型病態(tài)性的影響,提高模型參數(shù)的估計精度.由均方誤差的角度分析可知,TSVD通過引入少量偏差,降低方差,來實現(xiàn)均方誤差的下降,截斷參數(shù)則是改善模型參數(shù)估值均方誤差的關鍵因素.通過分析截掉奇異值后,TSVD模型參數(shù)估計方差與偏差的變化情況,提出了依據(jù)引入偏差量小于降低方差量確定截斷參數(shù)的方法,理論依據(jù)更為充分,可靠性與準確性更高.將采用新方法確定截斷參數(shù)的TSVD應用到測量坐標解算及PolInSAR植被高反演中,驗證了新方法的可行性和有效性,相比于GCV法和L曲線法,新方法確定的截斷參數(shù)有效提高了TSVD的解算質量,提高了坐標解算和植被高參數(shù)反演的精度和可靠性.
[Abstract]:The truncated singular value method (TSVD) improves the effect of model ill-condition by truncating the smaller singular value in the ill-conditioned matrix, and improves the estimation accuracy of model parameters. From the angle of mean square error, it can be seen that TSVD reduces the variance by introducing a small amount of deviation. The truncation parameter is the key factor to improve the estimation mean square error of model parameters. The method of determining the truncation parameter by introducing deviation is less than decreasing variance is put forward, and the theoretical basis is more sufficient. The reliability and accuracy of the new method are higher. TSVD, which uses the new method to determine the truncation parameters, is applied to the calculation of the measured coordinates and the inversion of the vegetation height of PolInSAR. The feasibility and validity of the new method are verified, compared with the GCV method and the L-curve method. The truncation parameters determined by the new method can effectively improve the quality of TSVD solution and improve the accuracy and reliability of coordinate calculation and vegetation high parameter inversion.
【作者單位】： 中南大學地球科學與信息物理學院;湖南科技大學地理空間信息技術國家地方聯(lián)合工程實驗室;
【基金】：國家自然科學基金項目(41531068;41474008;41574006;41674012)
【分類號】：Q948;TN957.52
，

本文編號：1694490