Coding method for multi-dimensional classification and recognition of shoe sole patterns

Abstract

The present invention proposes a coding method for multi-dimensional classification and recognition of shoe sole patterns, which includes the following steps: S1, screening a large number of clear and complete shoe sole sample patterns, selecting a sufficient number of unspecific shoe sole patterns for comparative analysis, and determining the analysis direction; S2, according to the analysis direction, summarize the semantic features of the sole pattern, extract the composition and feature dimensions of the sole pattern; enumerate the sample footprints, and repeatedly correct the features and dimensions until the characteristics are universal; S3, clarify the definition of sole pattern features and organizational rules , define the macro-features and micro-features through the characteristic components and dimensional relationships of the sole pattern, and encode the sole pattern features accordingly, and classify the sole patterns according to the coding rules.

Description

technical field [0001] The invention relates to the technical fields of computer graphics analysis and public security criminal science, in particular to a coding method for multi-dimensional classification and recognition of shoe sole patterns. Background technique [0002] Shoe prints (traces) are one of the most commonly encountered traces and physical evidence in case scene investigations, and they play an important role in many investigative methods such as scene reconstruction, perpetrator characteristics analysis, and parallel cases. In public security criminal investigation work, searching and comparing massive shoe print pictures is the key point of this technology. At present, the mainstream algorithm adopted in China is to convert images into binarized images and compare the similarity of binary images. However, because the on-site shoe prints are affected by the medium, carrier, extraction technology, and trace formation process, the on-site shoe print photos ha...

AI Technical Summary

Problems solved by technology

However, because the on-site shoe prints are affected by the medium, carrier, extraction technology, and trace formation process, the on-site shoe print photos have the characteristics of light spots, noise, low trace contrast, unevenness, even deformation, and overlapping, which seriously affect the binary image. The accuracy of the algorithm, and this problem is the objective performance of on-site shoe prints, the binary image algorithm cannot avoid this problem, and requires a lot of manual image correction, the effect is still not good, so many technicians gradually give up using this type of system comparison , and turned to manual sea selection of shoe prints, resulting in a lot of labor costs and low output

Images