ANonlinear PDEBasedEdgeDetection forImageMeasurement
Shujun Fu, Qiuqi Ruan · 2005
Imagemeasurement technique isanewdevel- oping methodinthenon-contact measurement andinstrument field, whereimageedgedetection isanimportant factor forthe measurement precision, especially inthecaseofblurry edges andadditional noise intheimage. We present anadaptive nonlinear PDEprocess, whichhastwosteps: first, weusea bidirectional flow(BDF)equation tosharpen andsmooth edges oftheimage, whereaninverse diffusion isperformed to sharpen edges along thenormal directions totheisophote lines, while anormaldiffusion isdonetoremovenoise andartifacts along thetangent directions; second, weusetheclassic differ- ential operator todetect edges. Experimental results onreal images demonstrate that, basedonsharpandsmoothedges, ouralgorithm canlocate edgesaccurately andthusmay promisingly produce moreexact measurement.