A step-by-step footprint recognition method combined with footwork features

Abstract

The invention provides a step-by-step footprint identification method combined with footwork features, which is characterized by comprising an off-line training process and an on-line identification process. The method comprises the steps of acquiring a step-by-step footprint image to be trained by the acquisition unit, and obtaining the initial step-by-step footprint image data set to be trained;extracting a single footprint from the footprint image to be trained; constructing a footwork energy map of the step-by-step footprint image to be trained; extracrting the step features of the footprint images to be trained for coarse-grained classification, and constructing a coarse-grained classification model according to the extracted step features; then constructing a fine-grained recognition model according to the energy map and the coarse-grained classification model. According to the principle of triangular stability, by using the weighted average of three single footprints to get thefootprint energy map as the discriminant image, the matching error is reduced by weighted average of footprints collected in different time periods, and the relative relationship between the left andright feet is maintained, which is the fusion of footprint characteristics and fully reflects the difference of footprints.

Description

technical field [0001] The invention relates to a recognition algorithm for a step footprint, in particular to a recognition algorithm for a step footprint combined with footwork features. Background technique [0002] Footprints are one of the important physical evidence with a relatively high rate of leaving at the crime scene. They play a very important role in criminal investigation cases. Generally, the footprints left at the crime scene include single footprints and multiple footprints. Through the analysis and inspection of these footprints, it can be inferred that a person The height, weight, age, and walking posture of the person can be analyzed, and the time left and the route of committing the crime can also be analyzed. When the conditions are good, it can also be tracked by features such as stride, and the suspect can be directly captured. Because everyone's height, weight, occupation, and habits are different, so everyone's gait characteristics are also differ...

AI Technical Summary

Problems solved by technology

[0006] The existing problems in the current research work on the footprints are: 1) The pressure change information needs to be collected dynamically, which is not suitable for the identification of the footprints at the scene of the case; 2) The step length and step width in the stride characteristics are relatively vulnerable The external environment is disturbed and unstable, which has certain limitations as the main research feature; 3) For the method of quantitative inspection of the stride characteristics, due to the instability of the stride characteristics, the result of the discrimination can only be obtained during the inspection process. With a relative range, the test results can only roughly exclude some suspects, narrowing the scope to a certain extent, which is of limited help to criminal investigators

Images