INSTITUTO ARGENTINO DE MATEMATICA ALBERTO CALDERON
Unidad Ejecutora - UE
Simultaneous Motion Detection and Background Reconstruction with a Conditional Mixed-State Markov Random Field
TOMÁS CRIVELLI; PATRICK BOUTHEMY; BRUNO CERNUSCHI FRÍAS; JIAN-FENG YAO
INTERNATIONAL JOURNAL OF COMPUTER VISION
Lugar: Springer Netherlands; Año: 2011 vol. 94 p. 295 - 295
In this work we present a new way of simultaneouslysolving the problems of motion detection and backgroundimage reconstruction. An accurate estimation of thebackground is only possible if we locate the moving objects.Meanwhile, a correct motion detection is achieved ifwe have a good available background model. The key of ourjoint approach is to define a single random process that cantake two types of values, instead of defining two differentprocesses, one symbolic (motion detection) and one numeric(background intensity estimation). It thus allows to exploitthe (spatio-temporal) interaction between a decision (motiondetection) and an estimation (intensity reconstruction)problem. Consequently, the meaning of solving both tasksjointly, is to obtain a single optimal estimate of such a process.The intrinsic interaction and simultaneity between bothproblems is shown to be better modeled within the so-calledmixed-state statistical framework, which is extended here toaccount for symbolic states and conditional random fields.