Pixel Statistics and False Alarm Area in Genetic Programming for Object Detection

Created by W.Langdon from gp-bibliography.bib Revision:1.4504

  author =       "Mengjie Zhang and Peter Andreae and Mark Pritchard",
  title =        "Pixel Statistics and False Alarm Area in Genetic
                 Programming for Object Detection",
  booktitle =    "Applications of Evolutionary Computing,
                 EvoWorkshops2003: Evo{BIO}, Evo{COP}, Evo{IASP},
                 Evo{MUSART}, Evo{ROB}, Evo{STIM}",
  year =         "2003",
  editor =       "G{\"u}nther R. Raidl and Stefano Cagnoni and 
                 Juan Jes\'us Romero Cardalda and David W. Corne and 
                 Jens Gottlieb and Agn\`es Guillot and Emma Hart and 
                 Colin G. Johnson and Elena Marchiori and Jean-Arcady Meyer and 
                 Martin Middendorf",
  volume =       "2611",
  series =       "LNCS",
  pages =        "455--466",
  address =      "University of Essex, UK",
  publisher_address = "Berlin",
  month =        "14-16 " # apr,
  organisation = "EvoNet",
  publisher =    "Springer-Verlag",
  email =        "mengjie@mcs.vuw.ac.nz",
  keywords =     "genetic algorithms, genetic programming, evolutionary
                 computation, applications, object recognition",
  isbn13 =       "978-3-540-00976-4",
  DOI =          "doi:10.1007/3-540-36605-9_42",
  abstract =     "This paper describes a domain independent approach to
                 the use of genetic programming for object detection
                 problems. Rather than using raw pixels or high level
                 domain specific features, this approach uses domain
                 independent statistical features as terminals in
                 genetic programming. Besides position invariant
                 statistics such as mean and standard deviation, this
                 approach also uses position dependent pixel statistics
                 such as moments and local region statistics as
                 terminals. Based on an existing fitness function which
                 uses linear combination of detection rate and false
                 alarm rate, we introduce a new measure called 'false
                 alarm area' to the fitness function. In addition to the
                 standard arithmetic operators, this approach also uses
                 a conditional operator ifin the function set. This
                 approach is tested on two object detection problems.
                 The experiments suggest that position dependent pixel
                 statistics computed from local (central) regions and
                 nonlinear condition functions are effective to object
                 detection problems. Fitness functions with false alarm
                 area can reflect the smoothness of evolved genetic
                 programs. This approach works well for the detecting
                 small regular multiple class objects on a relatively
                 uncluttered background.",
  notes =        "EvoWorkshops2003",

Genetic Programming entries for Mengjie Zhang Peter Andreae Mark Pritchard