| 摘要 |
PURPOSE:To improve convergence as compared with a conventional sequential approximating method and increase the likelihood of a solution by employing algorithm which uses the result of sequential approximation based upon non- dense matrix as the initial value of sequential approximation based upon a finer matrix. CONSTITUTION:Sequential approximation with matrix size 64X64 as shown in a figure (a) is performed before sequential approximation with matrix size 512X 512 shown in a figure (c). Then, sequential approximation with matrix size 128X 128 shown in a figure (b) is carried out and the sequential approximation with the matrix size 512-512 is performed lastly. The value of picture element data M4 (I, J) is divided by four and rounded to obtain the initial value of each picture element data M3 (I, J) of the 128X128 matrix from the picture element data M4 (I, J) of the 64X64 matrix. The M3 (I, J) is converged on the initial value, and sequential approximation of M2 (I, J) of the 256X256 matrix and lastly M (I, J) is carried out. |
