Handwritten recognition is the ability of the computer to read, recognize and interpret the various styles of handwritten from different sources such as papers, documents, photographs, stored screens and devices. An intelligent word recognition framework is developed with the help of optical character recognition (OCR). The features of the character written are recognized by the unique strokes. The proposed work is formulated based on (ARIMA) Auto regressive integrated moving average estimation model to determine the uniqueness present in it. The presented model calculates the time series update of the character is fetched to determine the pattern. Further the ARIMA model determines the prediction statistics based on unique statistical metrics arrived from the independent data. The proposed approach generates the lead lag parameter on independent inputs; further the maximum correlation pattern determines the character recognition. The input dataset is collected from real time camera, and training data from different sources of internet. To avoid human errors, automated recognition systems are derived in recent days. The real time camera enabled system, recognize the characters accurately using neural network model of auto regressive structure. Detection accuracy of 95% is achieved with the proposed system.

Handwritten recognition, signature verification, neural networks, pattern recognition, statistical metrics.