Footprint Recognition and Feature Extraction Method Based on Artificial Intelligence

摘要

In the modern society where artificial intelligence is gradually being used, identity recognition is widely used in various places. Footprint recognition is different from face recognition, its feature extraction is more difficult, and it is often used in related criminal investigation fields. In this study, Haar and Local Binary Pattern (LBP) are selected as the feature extraction methods. These two methods are used to extract the feature vectors of digging and tread marks respectively, and Adaptive Boosting (Adaboost) is selected as the classification algorithm, based on Haar features and LBP feature training respectively. Algorithms are developed and tested for detecting pick marks and tread marks. The test results show that for pick marks and tread marks, the artificial intelligence method using Haar and LBP features has a better detection effect.