作者: Douglas J Hunt , Loren W Nolte , W Howard Ruedger
DOI: 10.1016/0734-189X(88)90062-X
关键词:
摘要: Abstract A statistical signal detection theory approach can be used for curve in digital images corrupted by random noise. This is shown to a refinement of the frequently Hough transform, resulting improved performance, both deciding presence or absence image and determining location an existing curve. Detection performance evaluated using receiver operating characteristics (ROC) curves. Location estimation measured deriving equations transform probability correctly estimating The performances these two approaches are compared various signal-to-noise ratios found significantly different some values ratio.