Sign In to Follow Application
View All Documents & Correspondence

Scalable Context Aware Signature Recognition For Real Time Data Stream Classification

Abstract: A computer-implemented system and method for automatic signature verification are disclosed. Reference signature samples and a test signature are first subjected to preprocessing operations to obtain normalized representations, followed by segmentation and extraction of discriminative features. The extracted features are evaluated in parallel using a deep bidirectional long short-term memory–based sequence analysis module, a dynamic time warping–based comparison coupled with a support vector machine classifier, and a comparator-based support vector machine classifier. Outputs generated by the parallel classification modules are combined using a score fusion mechanism to produce a unified confidence measure. A decision-making module processes the fused score to classify the test signature as genuine or forged. By integrating multiple complementary comparison techniques within a single verification framework, the disclosed system improves robustness and accuracy of signature verification, particularly in the presence of intra-class variations and skilled forgeries.

Get Free WhatsApp Updates!
Notices, Deadlines & Correspondence

Patent Information

Application #
Filing Date
16 April 2026
Publication Number
22/2026
Publication Type
INA
Invention Field
COMPUTER SCIENCE
Status
Email
Parent Application

Applicants

Vinay Kumar
Department of CSIT Krishna Institute of Technology, Ghaziabad, India
Swasti Singhal
Department of CSIT, Krishna Institute of Engineering & Technology (KIET), Ghaziabad, Delhi-NCR, Uttar Pradesh, India
Dr. Prince Gupta
Department of CSIT, Krishna Institute of Engineering & Technology (KIET), Ghaziabad, Delhi-NCR, Uttar Pradesh, India
Dr. Rohit Vashisht
Department of CSIT, Krishna Institute of Engineering & Technology (KIET), Ghaziabad, Delhi-NCR, Uttar Pradesh, India
Dr. Ashima Arya
Department of CSIT, Krishna Institute of Engineering & Technology (KIET), Ghaziabad, Delhi-NCR, Uttar Pradesh, India
Dr. Shrankhla Saxena
Department of CSIT, Krishna Institute of Engineering & Technology (KIET), Ghaziabad, Delhi-NCR, Uttar Pradesh, India
Dr. Ankur Garg
Department of CSIT, Krishna Institute of Engineering & Technology (KIET), Ghaziabad, Delhi-NCR, Uttar Pradesh, India
Shivangi Tyagi
Department of CSIT, Krishna Institute of Engineering & Technology (KIET), Ghaziabad, Delhi-NCR, Uttar Pradesh, India
Satyam Shivam Sunderam
Department of CSIT, Krishna Institute of Engineering & Technology (KIET), Ghaziabad, Delhi-NCR, Uttar Pradesh, India
Dr. Sonia Deshmukh
Department of CSIT, Krishna Institute of Engineering & Technology (KIET), Ghaziabad, Delhi-NCR, Uttar Pradesh, India
Ms. Latika Sharma
Department of CSIT, Krishna Institute of Engineering & Technology (KIET), Ghaziabad, Delhi-NCR, Uttar Pradesh, India

Inventors

1. Swasti Singhal
Department of CSIT, Krishna Institute of Engineering & Technology (KIET), Ghaziabad, Delhi-NCR, Uttar Pradesh, India
2. Dr. Prince Gupta
Department of CSIT, Krishna Institute of Engineering & Technology (KIET), Ghaziabad, Delhi-NCR, Uttar Pradesh, India
3. Dr. Rohit Vashisht
Department of CSIT, Krishna Institute of Engineering & Technology (KIET), Ghaziabad, Delhi-NCR, Uttar Pradesh, India
4. Dr. Ashima Arya
Department of CSIT, Krishna Institute of Engineering & Technology (KIET), Ghaziabad, Delhi-NCR, Uttar Pradesh, India
5. Dr. Shrankhla Saxena
Department of CSIT, Krishna Institute of Engineering & Technology (KIET), Ghaziabad, Delhi-NCR, Uttar Pradesh, India
6. Dr. Ankur Garg
Department of CSIT, Krishna Institute of Engineering & Technology (KIET), Ghaziabad, Delhi-NCR, Uttar Pradesh, India
7. Shivangi Tyagi
Department of CSIT, Krishna Institute of Engineering & Technology (KIET), Ghaziabad, Delhi-NCR, Uttar Pradesh, India
8. Satyam Shivam Sunderam
Department of CSIT, Krishna Institute of Engineering & Technology (KIET), Ghaziabad, Delhi-NCR, Uttar Pradesh, India
9. Dr. Sonia Deshmukh
Department of CSIT, Krishna Institute of Engineering & Technology (KIET), Ghaziabad, Delhi-NCR, Uttar Pradesh, India
10. Ms. Latika Sharma
Department of CSIT, Krishna Institute of Engineering & Technology (KIET), Ghaziabad, Delhi-NCR, Uttar Pradesh, India
11. Mr Vinay Kumar
Department of CSIT, Krishna Institute of Engineering & Technology (KIET), Ghaziabad, Delhi-NCR, Uttar Pradesh, India

Claims

1. A mechanism for verifying signatures using multi-staged approaches involving preprocessing, segmentation, and feature extraction. 2. An approach that involves utilizing DBLSTM, DTW, and comparators to enable accurate signature matching. 3. A score fusion unit which helps to aggregate various similarities in order to enhance the performance of signature matching. 4. A decision making unit that uses the similarity scores in determining whether the signature is a forgery or genuine. 5. A computer readable medium comprising executable instructions for verifying and classifying signatures through comparisons.

Specification

Description:[0003] Various contemporary uses of handwritten or behavior-based signatures require efficient methods of verifying their legitimacy for proper authentication purposes to prevent any form of unauthorized access or fraud. The classification approach based on signatures finds application in domains such as computer and information security surveillance, fraud identification, network traffic examination, identification, and intelligence gathering operations where records or signatures to be verified can be matched to certain reference signatures.
[0004] Signature verification systems typically use a comparison between a signature being analyzed and one or several reference signatures to ascertain whether it is genuine or forged. This type of verification may apply to a variety of uses, including financial transaction verification, entry into restricted areas, documents validation, or digital identities management systems.
[0005] Single comparison approaches, as well as single-classifier-based solutions for signature verification are known to have some shortcomings as they do not necessarily consider all aspects of signatures that may include temporal, spatial and dynamic features. Moreover, known systems may demonstrate reduced accuracy when facing intra-class variability and well-trained forgeries, as well as differences associated with varying acquisition equipment and conditions.
5
[0006] In addition, any system that uses single decision-making process is vulnerable to the influence of noise or errors of segmentation or extraction of features, leading to false classification results. Therefore, the present invention aims at development of more advanced technology that will incorporate multiple methods of analysis, as well as classification and fusion of results produced by multiple classifiers. , Claims:1. A mechanism for verifying signatures using multi-staged approaches involving preprocessing, segmentation, and feature extraction. 2. An approach that involves utilizing DBLSTM, DTW, and comparators to enable accurate signature matching. 3. A score fusion unit which helps to aggregate various similarities in order to enhance the performance of signature matching. 4. A decision making unit that uses the similarity scores in determining whether the signature is a forgery or genuine. 5. A computer readable medium comprising executable instructions for verifying and classifying signatures through comparisons.

Documents