A. Pikaz et Ih. Dinstein, USING SIMPLE DECOMPOSITION FOR SMOOTHING AND FEATURE POINT DETECTION OF NOISY DIGITAL CURVES, IEEE transactions on pattern analysis and machine intelligence, 16(8), 1994, pp. 808-813
This correspondence presents an algorithm for smoothed polygonal appro
ximation of noisy digital planar curves, and feature point detection.
The, resulting smoothed polygonal representation preserves the signs o
f the curvature function of the curve. The algorithm is based on a sim
ple decomposition of noisy digital curves into a minimal number of con
vex and concave sections. The location of each separation point is opt
imized, yielding the minimal possible distance between the smoothed ap
proximation and the original curve. Curve points within a convex (conc
ave) section are discarded if their angle signs do not agree with the
section sign, and if the resulted deviations from the curve are less t
han a threshold epsilon which is derived automatically. Inflection poi
nts are curve points between pairs of convex-concave sections, and cus
ps are curve points between pairs of convex-convex or concave-concave
sections. Corners and points of local minimal curvature are detected b
y applying the algorithm to respective total curvature graphs. The det
ection of the feature points is based on properties of pairs of sectio
ns that are determined in an adaptive manner, rather than on propertie
s of single points that are based on a fixed-size neighborhood. The de
tection is therefore reliable and robust. Complexity analysis and expe
rimental results are presented.