| 摘要 |
PURPOSE:To reconstitute images at high speed and at low cost without degrading the reconstituted images so much with a little computation processing even when projecting directions are extremely decreased by multiplying projective data to an inverse pseudo matrix to always minimize the square of reconstituting error. CONSTITUTION:Models are set to the body of an object and projection (A) and the matrix is calculated to the object and projection models based on the density value influence relation between a sampling point and an arbitrary point and the relation position of projected beams (B.) Then, special value decomposition is performed to this object and projection model matrix, and the inverse pseudo matric to minimize the error of reconstitution is calculated beforehand (C.) Afterwards, the number of times for multiplying the projective data is reduced by multiplying a projective data vector to this inverse pseudo matrix or omitting the component of the inverse pseudo matrix close to zero. Therefore, when the body of the object is close to the model, the satisfactory images can be reconstituted even when the projecting directions are extremely decreased and even when the body of the object is not matched to the model, the model images are reconstituted so as to minimize the deviation. |
