Object Recognition Based on Smoothing Segmented-contour

史思琦,石光明,陈旭阳,李甫
DOI: https://doi.org/10.3788/gzxb20103905.0929
2010-01-01
Abstract:An object recognition algorithm based on segmented-smoothing contour is proposed.The contour is divided into two kinds of zones by curvature:feature zones and non-feature zones.Then,the different kinds of zones are smoothed with Gaussian functions with different variance.Finally,the object recognition algorithm based on affine moment-invariant is applied to recognize the smoothed contours.The simulation result shows the better contour-smoothing results and the significant improvement to recognition rate.
What problem does this paper attempt to address?