Recognition Of Hong Kong Dollar Based On Multi-Spectral Image

Based on multi-spectral paper image recognition is the hotspot of artificial intelligence and machine learning in recent years,the research of identification algorithm of banknote image is very important to maintain the security and stability of financial market in China.The analysis of multi-spectral images of Hong Kong through digital image pattern recognition technology can quickly and efficiently identify the denomination,orientation and version of the Hong Kong dollar and the Hong Kong dollar crown,and accurately identify the authenticity of the Hong Kong dollar.Therefore,the study of Hong Kong dollar identification algorithm will have important theoretical and practical value for the safe circulation of Hong Kong and the stability of China's economic market.This paper first analyzes the Hong Kong dollars of various versions and denominations,as well as the anti-counterfeiting features of Hong Kong dollar,and also performs geometric correction and preprocessing on Hong Kong dollar infrared reflection images.Secondly,this paper proposes an improved Haar feature extraction algorithm for extracting Hong Kong dollars Infrared reflection image features,and compare the template matching classifier on the Hong Kong dollar orientation and version recognition classification effect,complete the Hong Kong dollar orientation and version recognition work.Then we propose the localized image enhancement algorithm of Hong Kong dollar crown image,and use the template matching classifier and KNN classifier to classify the Hong Kong dollar characters to complete the Hong Kong dollar crown recognition.Finally,aiming at the shortcomings of the traditional feature block recognition algorithm of traditional Hong Kong dollar,this paper proposes a global feature recognition algorithm based on Tamura texture,which solves the problem of low robustness,slow computation and low recognition accuracy of traditional Hong Kong dollar pseudo-recognition algorithm problem.In this paper,we first analyze the Hong Kong dollars of various versions and denominations,as well as the anti-counterfeiting features of Hong Kong dollar and the geometric correction and preprocessing of Hong Kong dollar images.Secondly,this paper proposes an improved Harr feature extraction algorithm for extracting Hong Kong dollar infrared Image features,and compare the template matching and KNN classifier to Hong Kong dollar orientation and version recognition classification effect,complete the Hong Kong dollar orientation and version recognition work.And then use the SVM classifier to classify the characteristics of the Hong Kong dollar character grid and complete the recognition of the Hong Kong dollar crown.Finally,aiming at the shortcomings of the traditional feature recognition algorithm of traditional Hong Kong dollar,this paper proposes a global feature pseudometric algorithm based on Tamura texture,which is classified by SVM classifier,which solves the problem of low robustness and computational speed Slow and low recognition accuracy.