Signatures are seen everywhere – in Art, Cryptography, Music, Bank Cheques, etc. It is a mark of a person for a proof of identity and intent. It is distinctive and unique to that person. It’s also a way to confirm the person’s identity. Although it is unique and private to that person, it can be forged. Banks use the signature to verify the cheques, Cryptocurrencies use the digital signature to verify a transaction, Art museums use the signature to identify the art’s owner. Hence, Identifying the authenticity of the signature is of utmost importance.
In the past, signature validation has been done manually where the bank’s employee would authenticate the signature on the cheque, Art museum’s employee would authenticate the signature on the art. This requires a good eye to notice any subtle differences and is prone error. Also, it is quite a time consuming process. With advent of the latest technology with incredible capabilities, this process can be automated- to reduce the error and to fasten the process. In the 21st century where billions of transactions occur on a daily basis, it is imperative that this process is fast and seamless.
There are multiple ways to automate the process of signature validation. The traditional way is to obtain the image of the signature, use image processing techniques such as Gray Conversion, Noise reduction, Edge enhancement, Binarization and others to get the signature in a good format. Then, extract features from the processed image of the signature. High level features such as width, height, aspect ratio. Low level features from specific parts of the signature such as the count of the transition of black to white pixel and vice-versa. Then store these features to compare it with the signature that needs to be verified. While there are other approaches such as using Fuzzy models, Hidden Markov models and others, the current best approach is to use Deep Learning. Deep Learning has done wonders with images and it outperforms all the other approaches currently known.
The human resources required to process and verify the innumerable transactions that occur on a daily basis is no longer an option. The automation of signature validation is not just to verify but to detect fraud. Automated Signature validation is a solution for efficient and fast validation of the signature that is a must to offer the best to your customer.