The Biometrics system is getting popularity since last decade As per Information Technology industry demand. This techn-ology are satisfy authentication and authorization process needs. But the unimodal biometric system have own limitations. the limitation of unimodal, we can choosing the approach of multimodal biometric system. In this research paper choose the physiological model for face recognition and behavioural model for signature recognition. The recognition of face and signature used match score level fusion. In this fusion technology for secured authentication of person
In this article, we use pre-train VGG16, VGG19, and ResNet50 with ImageNet wights and our best CNN model to identify human fingerprint patterns. The system including pre-processing phase where the input fingerprint images first technique apply cropping and normalize for unwanted part remove of fingerprint images and normalize its dimension, second Image Enhancement for removing noise in to ridgelines, and last Canny Edge Detection technique for adjustment to smooth image with Gaussian to remove noise. Then apply one by one model on KVKR fingerprint dataset. Our best CNN model has automatically extracted features and RMSprop optimizer use for classification this features. This study performing experimental work of each pre-processed dataset and testing these three models with different dataset size of input train, test, and validation data. The VGG16 model got a better recognition accuracy than VGG19 and ResNet50 models.
In the previous decade, biometrics referred to the automatic recognition of persons based on their physiological or behavioural traits but unimodal biometrics have their limitations. Due to its potential to overcome some of the inherent limitations of single biometric modalities while simultaneously enhancing overall recognition rates, multimodal biometrics has recently gained prominence. In this research, we offer a multimodal biometric person authentication system based on pre-train transfer learning VGG16 with CNN and CNN models, that uses the user's face and fingerprint biometric traits. We have used the own collected samples of same person KVKR face and fingerprint dataset for experimental work. First, we have applied pre-processing data augmentation technique on face and fingerprint data then image enhancement techniques on fingerprint data. In the features extraction, we have extraction the hidden feature of the face and fingerprint images using pre-train VGG16 with CNN and CNN models. The hstack method has been used to combine the features and SoftMax classifier use for features classification. The fusion score is calculated using the fixed-rule-based maximum rule technique, finally we have done comparative analysis of the unimodal and multimodal biometric recognition system.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.