Anti-counterfeit Identification Research In The HK Banknotes Based On Multi-spectral Images

Identify the truth of the banknote is a hot research topic in the field of multi spectral images, based on the multi spectral image of HK security identification and recognition of belonging to the frontiers of the discipline, with important theoretical value and broad application prospect.Main problems faced by HK multi spectral image authentication algorithm is: HK version diversity, multi spectral image of the image features of different versions of HK $, anti-counterfeit point both many similarities, but not the same; circulation Hong Kong exist different degree of old and new, wear and tear, pollution condition. This has put forward a very high request in four stages of image preprocessing, face recognition and identification.In view of the above problems, we first versions of HK and the corresponding security features are analyzed, and improved edge detection algorithm. Secondly, this paper presents IR image of HK $, oriented and version identification algorithm based on, the algorithm combined with SOFM neural network, to HK $IR image as the object of study, extraction of HK $image grid gray feature, recognition of HK $oriented and version.Then, adopt the strategy of regional contrast, to put forward the multi spectral image region feature selection algorithm, and based on this proposed based on improved characteristics of effective information of gray, gray distribution features based on gray level co-occurrence matrix HK texture features extraction algorithm. Finally, in order to improve the classification ability, this paper also introduces the supervised training method, and proposes a classification algorithm based on Adaboost for the multi feature fusion of the Hong Kong dollar.. The problem of weak anti-interference ability, large computation workload and low recognition accuracy of the recognition of the currency in the recognition of the currency of the currency is solved.Through comparison with other banknote image pseudo Kam r ecognition algorithm have indicated that the algorithm proposed in this paper for HK version recognition accuracy rate higher, better adaptability; better adapted to the HK security features, for HK $false discriminating effect better, of lossless coins, mutilated coins and stronger anti-interference ability.