Readings on factorization methods for geometric structure and motion

Orthographic, Weak perspective, affine and para-perspective linear cameras

The classic paper that started the field is Tomasi and Kanade's factorization method for a linear cameras: (Unfortunaterly I couldn';t find the published versio one on-line, but here is a TR version) 
@article{bb37419,
        AUTHOR = "Tomasi, C. and Kanade, T.",
        TITLE = "Shape and Motion from Image Streams under Orthography:
A Factorization Method",
        JOURNAL = "IJCV",
        VOLUME = "9",
        YEAR = "1992",
        NUMBER = "2",
        MONTH = "November",
        PAGES = "137-154"}
Several extensions were published by Kanade's students, perhaps most notably, Weinshall and Tomasi's to a Weak perspective camera in PAMI95.

And Poelman and Kanade A Paraperspective Factorization Method for Shape and Motion Recovery. The TR version of the paper can be found here. 

@article{ poelman97paraperspective,
    author = "Conrad J. Poelman and Takeo Kanade",
    title = "A Paraperspective Factorization Method for Shape and Motion Recovery",
    journal = "IEEE Transactions on Pattern Analysis and Machine Intelligence",
    volume = "19",
    number = "3",
    pages = "206-218",
    year = "1997",
    url = "citeseer.nj.nec.com/poelman97paraperspective.html" }

Projective cameras

The scale factors in the homogeneous projection equation presents some additional problems when solving for projective structure. A brief introduction to this is given in Hartley and Zisserman, Ch 17.3.

One approach by Multi-Image Projective Structure and Motion determines these scale factors by first relating images in pairs and computing the fundamental matrix, then computing projective depths. An experimental comparison of different implementations of this algorithm performed by Bill Triggs is also available. 

@inproceedings{ ii96factorization,
    author = "Peter Sturm {II} and Bill Triggs",
    title = "A Factorization Based Algorithm for Multi-Image Projective Structure and Motion",
    booktitle = "{ECCV} (2)",
    pages = "709-720",
    year = "1996",
    url = "citeseer.nj.nec.com/sturm96factorization.html" }
Another approach has been developed by the Lund group through several papers. A good reading order of these are: Third, Rother and Carlsson have a method based on tranforming with the infinite homography. Linear Multi View Reconstruction and Camera Recovery