Fingerprint pre-processing and feature engineering to enhance agricultural products categorization

Abstract

Among many biometric technologies, fingerprint identification is the earliest, most pervasively applied, and cheapest biometric technology. It seamlessly corporates integrates sensors, biotechnology, electronic technology, digital image processing and pattern recognition. Fingerprint based identity authentication has expanded from the traditional criminal investigation to e-commerce, attendance, access control, credit card, endowment insurance system and other fields, and has become the popular biometric technology. As for the existing matching algorithm such as feature-point-based or image-based and so on, the accuracy and speed of fingerprint automatic identification are far from the satisfaction of many practical applications. In this work, we investigate the fingerprint feature extraction, and optimize the performance of fingerprint login authentication system by introducing a new feature extractor based on binary image through a brief thinning process. Aiming at the problems of low flexibility, high cost and low efficiency of algorithm in embedded chip of current fingerprint identification system, the automatic fingerprint identification system is comprehensively analyzed. Our proposed method can improve the design and optimization of the automatic fingerprint identification system. Our method can improve the accuracy of fingerprint image recognition in the automatic fingerprint identification system. Meanwhile, it can also improve the adaptability and stability of the automatic fingerprint identification system in various domains. Also, our method can substantially improve the feature extraction of agriculture products. Thereby, the classification task can be enhanced accordingly.