Research on Similarity Scores Calculation of Two-dimensional Bare Footprint Contour
As an important feature of bare footprints,contour plays an important role in the identification of same source and non-matches bare footprints.Aiming at the problem that the contour features of bare footprints are difficult to quantify,a similarity score calculation method is proposed to improve the scienti-ficity and accuracy of identification.Firstly,2120 bare footprint samples from 920 volunteers were col-lected,and the contour similarity of toe region,anterior margin and heel region were calculated by shape context algorithm after image preprocessing.The weights of three regions were normalized to 0.41,0.42 and 0.17 with the help of variable importance for the projection in partial least squares discriminant anal-ysis.After weighting the three regions,the overall contour similarity was calculated.Under the overall contour similarity,the same source and non-matches bare footprints were significantly different,and the accuracy rates of pattern recognition were 83.0％and 100.0％,respectively.Finally,a bare footprint similarity evaluation system was established,which could be used to calculate the similarity between the examinations and the samples.This system could realize the effective distinction between same source and non-matches bare footprints,which may assist the identification in real cases.
bare footprintcontour similarity scoreshape contextvariable importance for the projec-tionpattern recognition
NETL
NSTL
万方数据
