Study On Iris Template Protection Method Based On Improved Logistic Chaotic Mapping

As a type of biometric feature,iris is widely used for identification due to its uniqueness and other properties,but the leakage of iris identity templates gives an opportunity to unscrupulous elements,which in turn affects the security of user information.Therefore,the protection of iris original feature templates has become one of the main research directions in the field of iris recognition.This thesis focuses on the iris recognition template protection problem,using three iris template protection methods based on improved Logistic chaotic mapping.In the image preprocessing stage,all three methods use calculus operators to locate the iris,which is normalized using the Rubber Sheet model.Then the Log-Gabor transform is used to extract iris features and encode them into iris feature codes（Iris Code）.Method 1 adopts the iris template protection method based on composite Logistic chaos mapping and heterogeneous or,which firstly transforms the Iris Code and inputs it to the SHA-256 function to get the message digest,maps the message digest to the initial values and parameters of the composite Logistic chaos mapping as the key of the encryption system,generates a binary pseudo-random sequence after iteration and mapping,and then The pseudo-random sequence and the original iris feature sequence are iterated and mapped to generate the final encrypted iris template;Method 2 uses the iris template protection method based on composite Logistic chaos mapping and DNA encoding,firstly,the message digest of Iris Code is also transformed into the initial values and parameters of the composite Logistic chaos mapping,and the pseudo-random sequence is generated by iterative mapping,and then the Then,the average value of Iris Code is mapped to DNA coding and decoding parameters,and the pseudo-random sequence and the original iris feature sequence are DNA coded and decoded according to the parameter correspondence rule,and the decoding result is used as an encrypted template;Method 3 uses a three-dimensional Logistic chaos mapping and DNA coding based iris template protection method,which,on the basis of Method 2,extends the composite The third method is based on the second method,which extends the composite Logistic chaos mapping to three dimensions,and combines the three-dimensional Logistic chaos mapping with DNA encoding to achieve template encryption.Finally,all the encrypted templates generated by the three methods use Hamming distance for identity matching.In order to test the performance of the methods in this thesis,two sub-banks,CASIA Iris Image Database version 1.0 and CASIA-Iris V3-Interval,from the CASIA iris dataset were used for experimental simulations to test and analyze the recognition performance and security performance of the above three methods.The experimental results show that the correct recognition rates of method 1 are 98.97%and 96.81%,method 2 is 99.06%and97.89%,and method 3 is 99.18%and 98.64%in the two sub-databases,respectively.In terms of template encryption performance,the key spaces of the three methods are 2¹⁰⁶,24×2¹⁰⁶ and24×2³¹⁸,respectively,which are improved compared with other methods commonly used in the same field,and all of them satisfy non-linkability,irreversibility,and revocability,which can protect the security of iris templates while ensuring the correct recognition rate.