IS = { zkontrolovano 27 Jul 2010 },
  UPDATE = { 2010-07-27 },
  author = { {\v C}ech, Jan and Matas, Ji{\v r}{\' i} and
             Per{\v d}och, Michal},
  title = {Efficient Sequential Correspondence Selection by Cosegmentation},
  journal = { IEEE Transactions Pattern Analysis and Machine Intelligence},
  year = {2010},
  volume = {32},
  number = {9},
  month = {September},
  pages = {1568--1581},
  project = {1ET101210406, FP6-IST-027113, ICT-215078 DIPLECS, MSM6840770038},
  keywords = {correspondence, verification, dense stereo, growing,
              sequential decison, learning, SVM, wide-baseline  stereo,
              image retrieval, SIFT, RANSAC},
  annote = { In many retrieval, object recognition and wide baseline
    stereo methods, correspondences of interest points (distinguished
    regions) are commonly established by matching compact descriptors
    such as SIFTs. We show that a subsequent cosegmentation process
    coupled with a quasi-optimal sequential decision process leads to
    a correspondence verification procedure that (i) has high
    precision (is highly discriminative) (ii) has good recall and
    (iii) is fast.  The sequential decision on the correctness of a
    correspondence is based on simple statistics of a modified dense
    stereo matching algorithm. The statistics are projected on a
    prominent discriminative direction by SVM. Wald's sequential
    probability ratio test is performed on the SVM projection computed
    on progressively larger cosegmented regions.We show experimentally
    that the proposed Sequential Correspondence Verification (SCV)
    algorithm significantly outperforms the standard correspondence
    selection method based on SIFT distance ratios on challenging
    matching problems.},
  psurl = {http://doi.ieeecomputersociety.org/10.1109/TPAMI.2009.176},
  issn = {0162-8828},
  publisher = {IEEE Computer Society},
  address = {New York, USA},
  authorship = {50-30-20},