Recursive Factorization Method for the Paraperspective Model Based on the Perspective Projection

The factorization method, which allows us to reconstruct the motion of the camera and shape of the object simultaneously from multiple images, provides high stability in numerical computations and satisfactory results. To apply this method to real-time processing, the recursive factorization method has been proposed. However, factorization method based on the affine projection has a limitation in reconstruction accuracy, and to achieve accurate reconstruction, the motion should be restricted. To overcome this problem, we present a recursive factorization method for the paraperspective model based on the perspective projection. The present method is far superior to other ones, in that it not only achieves accurate Euclidean reconstruction in a short time but also provides high stability in numerical computations. Moreover, the method produces stable reconstruction in almost all cases even if some images contain errors because all images are treated as uniformly as possible.

• Type: Conference
• Year: 2000
• Where: ICPR
• Authors: Jun Fujiki, Takeshi Kurata
• Keywords: Accurate Euclidean Reconstruction, Computer Vision, ICPR 2000, Present Method, Recursive Factorization Method