Research Of High Quality Extraction Algorithm On Overlapped Fingerprint Based On Independent Component Analysis

With the development and application of biometric technology,fingerprint recognition technology is continuously popularized and used in real life and security protection due to its uniqueness and safety and reliability.It plays an extremely important role in the field of case detection and criminal investigation.Aiming at the separation of overlapping fingerprint traces left over from crime scenes,a high-quality extraction algorithm for overlapping fingerprints based on independent component analysis is studied.Independent component analysis is one of the main methods to solve blind source separation,When the source signal and the mixing method of the source signal are both unknown,the separation matrix of the mixed signal can be obtained by using only the non-Gaussian nature of the source signal and its independent basic assumptions,and the separation matrix of the mixed signal can be mixed.The independent components in the signal are separated to achieve blind source separation.Since its rise,independent component analysis has achieved vigorous development in many fields such as speech image processing,ultrasonic detection,and biomedical signals.Independent component analysis is one of the main methods to solve the blind source separation,when the source signal and the mixing mode of the source signal are unknown,the separation matrix of the mixed signal can be obtained only by using the non-Gaussianity of the source signal and the basic assumptions of its mutual independence,and then the independent in the mixed signal can be separated by using the separation matrix,so as to realize the blind source separation.In this paper,based on independent component analysis,the Fast ICA algorithm based on negative entropy maximization is used to process overlapping fingerprints,and the separation of overlapping fingerprints is realized.In this paper,based on independent component analysis,a fixed point algorithm based on maximum negative entropy is used to process overlapping fingerprints and Realize the separation of overlapping fingerprints.And introduce an improved arithmetic average Newton method with fifth-order convergence to improve the iterative formula of the fixed point algorithm(Fast ICA).Experimental results show that compared with the original Fast ICA algorithm,the improved algorithm has reduced the number of iterations,significantly accelerated the convergence speed,and greatly improved the efficiency.At the same time,it can ensure the fit between the separated single fingerprint and the source fingerprint.Considering that in actual situations,overlapping fingerprint images may have a lot of Gaussian and salt and pepper noise,this paper compares the denoising performance of several commonly used denoising algorithms,and finally determines that the adaptive fractional Alexander function(A-FAP)image denoising algorithm is used to denoise the noisy overlapping fingerprint images,and then the algorithm is improved The Fast ICA algorithm combined to achieve the separation of noisy overlapping fingerprint images.The experimental results show that first denoising the overlapping fingerprint image and then using the improved Fast ICA algorithm to separate the overlapping fingerprints,it has more advantages in preserving the image texture details and edge information,and the separation effect is better.