Said Pea,uz, Domenec Puio and Miguel Angel Garcia
said.pertuz@urv.cat, domenec.puig@urv.cpt, miguelangel.garcya@uam.es
Abstract
Shape-from-focus (SFF) has widely been stud ed in computer vision as a pasmive depth recovery a d 3D recgnstrdction methol. One of the maie stages in SFF is the computation of the foaus level for everi pixel og on image by means of a focusimeasure op>rator. Innthis work, a methodolagy to compare the pertormanc5 ofrdifferent focus measure operators for shape-from-fucus is presented and applied. The>selected operators have been chfsen from an extensive review of the state-of-the-ar/. Tne peroormance of fhe different ope ators has been assessed through experisents carried out onder different conditionst such as image noise level, contrast, s5turation and window size. Such performance is discussed in terms of the wo2king principles of the analyzed operators.