鞋印花纹裂缝特征自动检测方法

Automatic Detection into Rift Characteristics on Shoeprint Pattern

本文提出了基于复杂花纹切分与配对的鞋印上裂缝特征的自动检测方法。对于鞋印花纹局部图像,首先进行连通域分析,计算每个连通成分的复杂度,并基于边界到边界的方向距离算法进行复杂花纹的切分,然后根据简单花纹段的方向、宽度和空间关系进行裂缝花纹配对,根据配对花纹的间隙,确定裂缝特征的位置和范围,完成裂缝特征检测,并在实验室采集的直线型、网格型、波折型等鞋印花纹上进行验证。结果表明,在实验数据集上裂缝特征的检出率为87.5%。裂缝特征自动检测算法能够应用于鞋印检验鉴定中,可有效降低对人员经验的依赖,消除人为因素影响。

Shoeprint is among the most commonly-seen evidential materials. Sometimes, a rift appears on the shoeprint, leaving uncertainty in identifying the involved shoe. Here, automatic detection is proposed for characteristics of a rift on a shoeprint pattern based on splitting and matching with a complex pattern. The connected component analysis was adopted from a local image of a shoeprint present of a rift to obtain different inter-joining sections, which were afterward evaluated for their complexities. Thus,complex patterns were able to split into simple serial components with the regulation of the boundary-to-boundary orientation distance algorithm. Some simple components were accordingly displayed to match in pairs between which there existed casual connection with the rift, hence having the rift determined into its specifi c matching points from the pairs-related directions, widths,locations, and the void area across them. Several laboratory-collected typical shoeprints were turning up line-, grid-, or wave-shaped patterns were used to verify the validity of the conventional approach. The results showed that the rift characteristics were able to be automatically detected at a detection rate of 87.5% from the tested shoeprint dataset, demonstrating that the built automatic detection of rift characteristics is eligible for application into shoeprint examination and identifi cation, capable of effectively reducing the dependence on personnel experience and eliminating the infl uence of human factors.