Iris, fingerprint and DNA are the most prominent technology specifically used for personal identification purposes. The iris recognition is not so efficient due to the wavering morphological variations of devices. It leads to a huge challenge for the conventional processing methods like cognitive and statistical learning during the identification process. The availability of dataset and classification approaches makes a barrier to fulfilling the learning requirements under the learning framework. This research considers multi-level feature fusion concept and three different datasets like iris, fingerprint, and DNA datasets are considered for validation thus, this research concentrates on modeling a hierarchical method to measure both the spatial and temporal features using Convolutional Neural Networks and Long Short Term Memory (CNN-LSTM) to predict the personal identification. The proposed hierarchical model works for processing the image at the initial stage and recognition/prediction model by transforming the image into a recognition label. The prediction module is based on a statistical learning process that includes ReLU, different input gates of LSTM, and Dense with ReLU. The dataset is collected from the online available Kaggle repository, and the experimentation is validated with various prevailing methods in terms of average training and validation accuracy. This hybrid model outperforms the prevailing methods and shows superiority to other models.

Iris recognition, DNA recognition, morphological variation, spatial and temporal features, convolutional neural network, long short-term memory