基于平面赤足足迹轮廓的相似度计算研究

Research on Similarity Scores Calculation of Two-dimensional Bare Footprint Contour

轮廓作为赤足足迹的一个重要特征,在同源与异源赤足足迹的检验鉴定中发挥着重要作用。针对赤足足迹的轮廓特征难以量化的问题提出一种相似度计算方法,以提高检验鉴定的科学性和准确性。首先收集920名志愿者的2120枚赤足足迹样本,经过图像预处理后利用形状上下文算法对赤足足迹脚趾区、跖前缘和跟后缘的轮廓相似度进行计算。借助偏最小二乘判别分析计算三个区域的变量投影重要性,经权重归一化后分别为0.41、0.42和0.17。将3个区域设置权重后计算总体轮廓相似度,在该相似度下同源、异源赤足足迹具有显著差异且模式识别正确率分别为83.0%和100.0%。最终建立了一个赤足足迹相似度评估体系,该体系可对赤足足迹检材与样本之间的相似度进行计算,实现对于同源与异源赤足足迹的有效区分,辅助检验鉴定工作。

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 scientificity and accuracy of identification.Firstly,2120 bare footprint samples from 920 volunteers were collected,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 analysis.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.