A system for forensic face sketch generation, enhancement, and recognition, and a method thereof

IN605227BActive Publication Date: 2026-10-07SRM UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
IN202541048276
Authority / Receiving Office
IN · IN
Patent Type
Patents
Current Assignee / Owner
Filing Date
2025-05-19
Publication Date
2026-10-07

AI Technical Summary

Technical Problem

Current forensic face recognition systems in India rely heavily on manual sketching, which is slow and subjective, and existing digital tools often produce unrealistic sketches, while AI-based systems cannot process hand-drawn images, leading to inefficiencies and security concerns.

Method used

A unified system integrating sketch creation, enhancement, and recognition, utilizing a user-friendly interface, deep learning for image processing, and cloud-based storage for secure, efficient forensic face identification.

Benefits of technology

The system automates the process from sketch creation to suspect matching, reducing investigation time and improving accuracy by generating high-quality, photo-realistic images for effective suspect identification.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 00000030_0000
    Figure 00000030_0000
Patent Text Reader

Abstract

ABSTRACT A SYSTEM FOR FORENSIC FACE SKETCH GENERATION, ENHANCEMENT, AND RECOGNITION, AND A METHOD THEREOF The present disclosure discloses a system for forensic face sketch generation, enhancement, and recognition, and a method thereof. The system (100) comprises a user interface (UI) module (102) operable on a remote device (104) with a secure user authentication sub-module (106). A sketch construction module (110) enables users to create composite sketches using predefined facial components or upload external hand-drawn images. The system (100) further includes an image enhancement module (112) utilizing a PyTorch-based GAN model with a U-Net generator and PatchGAN discriminator, trained on CUHK and Indian Face Sketch datasets, to convert sketches into photorealistic images. An image storage and cloud indexing module (114) stores and indexes enhanced images on a cloud service. A recognition module (116) compares input images against stored faces, displaying similarity scores and matched results. The system facilitates real-time, accurate, and secure forensic identification, enhancing the effectiveness of law enforcement investigations.
Need to check novelty before this filing date? Find Prior Art

Description

FIELDThe present disclosure generally relates to the field of computer vision and forensictechnology. More particularly, the present disclosure relates to a system for forensicface sketch generation, enhancement, and recognition, and a method thereof.DEFINITIONAs used in the present disclosure, the following terms are generally intended to havethe meaning as set forth below, except to the extent that the context in which theyare used indicates otherwise.The term "one-time password (OTP)" refers to an automatically generated numericor alphanumeric string of characters that authenticates a user for a single transactionor login session.The term "application program interface (API)" refers to a set of rules and protocolsthat allow different software applications to communicate and share data.The term "Secure Hash Algorithm (SHA) 256-bit" refers to a specific cryptographichash function that produces a 256-bit hash value. This algorithm is part of the SHA2 family and is used to verify the integrity of data, ensuring it hasn't been tamperedwith.The term "Generative Adversarial Network (GAN)" refers to a deep learningarchitecture. It trains two neural networks to compete against each other to generatemore authentic new data from a given training dataset.The term "Contrast Limited Adaptive Histogram Equalization (CLAHE)" refers toa technique used in image processing to enhance contrast and reduce noise inimages, particularly those with low contrast or high noise.The term "graphical user interface (GUI)" refers to a form of user interface thatallows users to interact with electronic devices through graphical icons and visualindicators such as secondary notation.The above definitions are in addition to those expressed in the art.BACKGROUNDThe background information herein below relates to the present disclosure but isnot necessarily prior art.In India, forensic face recognition heavily relies on manual sketching byprofessional artists, which is slow and often unreliable, as it depends on the artist'sskill and the witness's memory. Although several state-level law enforcementagencies have adopted facial recognition technologies, they predominantly focuson photo-based identification and do not accommodate inputs derived fromeyewitness sketches. As a result, such technologies are ineffective in cases wherephotographic evidence is unavailable. National systems like the NationalAutomated Facial Recognition System (NAFRS) don't support sketch-basedmatching and raise privacy concerns. Because of this gap, identifying suspectsquickly and accurately remains a major challenge for law enforcement.Existing solutions in the field typically offer siloed functionalities, focusing eitheron sketch creation or face recognition, but rarely both. Manual sketches, althoughartistic, are dependent on skilled personnel and lack consistency across cases. Somedigital sketching tools have emerged to address this issue, but often generatesimplistic or cartoon-like outputs that are visually unrealistic. Additionally, whileface recognition systems powered by AI or cloud platforms do offer robustidentification from photographs. Academic efforts such as CLIP4Sketch haverecently demonstrated the potential of sketch-to-photo AI matching, reinforcing theopportunity to build more comprehensive systems.Despite their advancements, traditional systems face critical limitations, primarilydue to their lack of integration. Most tools focus either on sketch generation or facerecognition, but not on a unified workflow. Manual sketching depends onprofessional artists, making it slow and subjective. Digital sketch tools oftenproduce unrealistic images, reducing the chances of accurate identification. Facerecognition systems, while effective with photographs, cannot process hand-drawnor composite sketches. Security is another concern many systems rely on basicpassword protection without multi-factor authentication. Additionally, storing dataonly on local machines limits scalability and increases the risk of data loss or systemfailure.Therefore, there is felt a need for a system for forensic face sketch generation,enhancement, and recognition, and a method thereof that alleviates theaforementioned drawbacks.OBJECTSSome of the objects of the present disclosure, which at least one embodiment hereinsatisfies, are as follows:It is an object of the present disclosure to ameliorate one or more problems of theprior art or to at least provide a useful alternative.An object of the present disclosure is to provide a system for forensic face sketchgeneration, enhancement, and recognition, and a method thereof.Another object of the present disclosure is to provide a system that enables end-toend forensic processing by integrating sketch creation, image enhancement, andface recognition into a single unified platform.Still another object of the present disclosure is to provide a system that has a userfriendly interface.Yet another object of the present disclosure is to provide a system that offers cloudscalability and reliability.Still another object of the present disclosure is to provide a system that reducesinvestigation time by automating the entire process from sketch creation to suspectmatching.Yet another object of the present disclosure is to provide a system that allowsstraightforward expansion, data security, and remote access from different crimeunits anywhere.Still another object of the present disclosure is to provide a system that offers robustauthentication and data protection.Yet another object of the present disclosure is to provide a system that is flexibleand easy to deploy.Still another object of the present disclosure is to provide a system that increasesrecognition success by generating high-quality images that better match real-worldsuspect databases.Another object of the present disclosure is to provide a system that significantlyreduces investigation time.Still another object of the present disclosure is to provide a system that createscomposite facial sketches based on verbal eyewitness accounts.Yet another object of the present disclosure is to provide a system that enhancesrough sketches into realistic, photo-like images for better visual clarity.Still another object of the present disclosure is to provide a method for end-to-endforensic face identification through a composite sketch image.Other objects and advantages of the present disclosure will be more apparent fromthe following description, which is not intended to limit the scope of the presentdisclosure.SUMMARYThe present disclosure envisages a system for forensic face sketch generation,enhancement, and recognition, and a method thereof.The system includes a user interface (UI) module, a remote device, a userauthentication sub-module, a database, a sketch construction module, an imageenhancement module, an image storage and cloud indexing module, and arecognition module.The user interface (UI) module operable on a remote device, wherein the UI modulecomprises a user authentication sub-module configured to receive login credentialsfrom an authorized user, verify the credentials against hashed values stored in adatabase, in a first mode, dispatch an on-time password (OTP) via an emailapplication program interface (API) to a registered email of the user for verificationthrough a sketch construction login interface and in a second mode, validatecredentials hashed with SHA-256 (Secure Hash Algorithm 256-bit) for access viaa recognition login interface.The sketch construction module operatively connected to the user authenticationsub-module and comprises a graphical user interface (GUI) built using JavaFX,configured to allow the user to construct a composite face sketch by selecting,dragging, and dropping facial components selected from a predefined componentset, including at least eyes, noses, lips, head shapes, eyebrows, and identificationmarks, provide resizing, repositioning, and rotation functionalities using boundingbox controls, optionally accept external hand-drawn sketch images for integration,and generate a digital sketch image as output.The image enhancement module is operatively coupled to the sketch constructionmodule and configured to receive the digital sketch image from the sketchconstruction module.In an embodiment, the image enhancement module comprises a deep learningmodel implemented by a PyTorch-based Generative Adversarial Network (GAN),the GAN comprising a U-Net generator and a PatchGAN discriminator trained onthe Chinese University of Hong Kong (CUHK) and Indian Face Sketch datasets,and an image processing engine implementing an OpenGV configured to applycolor corrections, brightness, adjustment, and sharpening operations, wherein theenhancement module outputs a photorealistic face image.The image storage and cloud indexing module is operatively connected to the imageenhancement module and configured to store the photorealistic face image in acloud-based storage service, dynamically index the stored image into the cloudbased storage service for face recognition purposes, and provide the indexed imageas an input for recognition.The recognition module is functionally integrated with the image storage and cloudindexing module, and comprises a GUI built using Java Swing for user interaction,configured to receive as input an image selected by the user, wherein the inputimage includes a freshly created sketch, a GAN-enhanced image, or an externalhand-drawn sketch, perform a face matching operation using the cloud-basedstorage service to compare the input image against indexed faces and computesimilarity scores, display the matched image with similarity score if the similarityscore is above a predefined threshold, and otherwise display a "No Match Found".In an embodiment, the system enables real-time sketch creation, enhancement,cloud-based storage, and face recognition to assist in forensic investigations.The present disclosure further envisages a method for end-to-end forensic faceidentification through a composite sketch image. The method includes thefollowing steps:- authenticating, by a user interface module, a user via credential hashingand OTP validation;- constructing, by a sketch construction module, a suspect facial sketchusing a drag-and-drop GUI that enables selection and manipulation offacial components;- enhancing, by an image enhancement module, the constructed sketchusing a Generative Adversarial Network (GAN) model, wherein the GANincludes a U-Net-based generator and a PatchGAN-based discriminator;- storing, by an image storage and cloud indexing module, the enhancedimage in a cloud-based storage service;- indexing, by an image storage and cloud indexing module, the storedimage into the cloud-based storage service;- comparing, by a recognition module, an uploaded image with images inthe cloud-based storage service using similarity thresholds; and- displaying on a GUI the matching result with a similarity percentage if amatch is found, or displaying a "No Match Found" message otherwise.BRIEF DESCRIPTION OF THE ACCOMPANYING DRAWINGA system for forensic face sketch generation, enhancement, and recognition, and amethod thereof of the present disclosure will now be described with the help of theaccompanying drawing, in which:Figure 1 illustrates a block diagram of a system for forensic face sketch generation,enhancement, and recognition, and a method thereof in accordance with the presentdisclosure;Figures 2A and 2B illustrate a flow chart depicting the steps involved in a methodfor end-to-end forensic face identification through a composite sketch image inaccordance with an embodiment of the present disclosure;Figure 3 illustrates the operational flow of the system for forensic face sketchgeneration, enhancement, and recognition in accordance with the presentdisclosure;Figure 4 illustrates the sketch construction flow in accordance with the presentdisclosure;Figure 5 illustrates the recognition flow in accordance with the present disclosure;Figures 6A-6F illustrate the user interface (UI) sketch construction of the two-stepverification, sketch construction canvas page, drag and drop features, sketch to facegeneration (enhancement), and recognition scenes in accordance with the presentdisclosure; andFigure 7 illustrates the recognition scene of the hand-drawn sketch in accordancewith the present disclosure.LIST OF REFERENCE NUMERALS100 - System102 - User Interface (UI) Module104 - Remote Device106 - User Authentication Sub-Module106a - Sketch Construction Login Interface106b - Recognition Login Interface108 - Database110 - Sketch Construction Module110a - Graphical User Interface (GUI)112 - Image Enhancement Module114 - Image Storage and Cloud Indexing Module116 - Recognition ModuleDETAILED DESCRIPTIONEmbodiments, of the present disclosure, will now be described with reference tothe accompanying drawing.Embodiments are provided so as to thoroughly and fully convey the scope of thepresent disclosure to the person skilled in the art. Numerous details are set forth,relating to specific components, and methods, to provide a complete understandingof embodiments of the present disclosure. It will be apparent to the person skilledin the art that the details provided in the embodiments should not be construed tolimit the scope of the present disclosure. In some embodiments, well-knownprocesses, well-known apparatus structures, and well-known techniques are notdescribed in detail.The terminology used, in the present disclosure, is only for the purpose ofexplaining a particular embodiment and such terminology shall not be consideredto limit the scope of the present disclosure. As used in the present disclosure, theforms "a," "an," and "the" may be intended to include the plural forms as well,unless the context clearly suggests otherwise. The terms "including," and "having,"are open ended transitional phrases and therefore specify the presence of statedfeatures, integers, steps, operations, elements, and / or components, but do not forbidthe presence or addition of one or more other features, integers, steps, operations,elements, components, and / or groups thereof. The particular order of stepsdisclosed in the method and process of the present disclosure is not to be construedas necessarily requiring their performance as described or illustrated. It is also to beunderstood that additional or alternative steps may be employed.When an element is referred to as being "engaged to," "connected to," or "coupledto" another element, it may be directly engaged, connected, or coupled to the otherelement. As used herein, the term "and / or" includes any and all combinations ofone or more of the associated listed elements.Forensic face recognition in India still relies on manual sketching, which is slow,subjective, and inconsistent. Existing systems mostly support photo-basedidentification and cannot match hand-drawn sketches. Current tools are fragmented,offering either sketching or recognition, but not both. Digital sketch tools oftenproduce unrealistic results, and AI-based recognition systems can't processsketches. Additionally, most platforms lack strong security and rely on localstorage, limiting scalability and reliability. These gaps make suspect identificationslow and less effective.To address the issues of the existing systems and methods, the present disclosureenvisages a system for forensic face sketch generation, enhancement, andrecognition (hereinafter referred to as "system 100") and a method (hereinafterreferred to as "method 200") for end-to-end forensic face identification through acomposite sketch image. The system 100 will now be described with reference toFigure 1, and method 200 will be described with reference to Figures 2A and 2 B.Referring to Figure 1, the system 100 comprises a user interface (UI) module 102,a remote device 104, a user authentication sub-module 106, a database 108, a sketchconstruction module 110, an image enhancement module 112, an image storage andcloud indexing module 114, and a recognition module 116.The user interface (UI) module 102 operable on the remote device 104, wherein theUI module 102 comprises the user authentication sub-module 106 configured toreceive login credentials from an authorized user, verify the credentials againsthashed values stored in the database 108, in a first mode, dispatch an on-timepassword (OTP) via an email application program interface (API) to a registeredemail of the user for verification through a sketch construction login interface 106aand in a second mode, validate credentials hashed with SHA-256 (Secure HashAlgorithm 256-bit) for access via a recognition login interface 106b.In an embodiment, the database is a MySQL database used to store credentials forboth sketch construction and recognition modules, hashed using BCrypt and SHA256, respectively, and the OTP in the sketch construction login interface 106a istransmitted using Jakarta Mail API, forming a two-factor authentication mechanismintegrated with hashed password validation.The sketch construction module 110 operatively connected to the userauthentication sub-module 106 and comprises a graphical user interface (GUI) 110abuilt using JavaFX, configured to allow the user to construct a composite facesketch by selecting, dragging, and dropping facial components selected from apredefined component set, including at least eyes, noses, lips, head shapes,eyebrows, and identification marks, provide resizing, repositioning, and rotationfunctionalities using bounding box controls, optionally accept external hand-drawnsketch images for integration, and generate a digital sketch image as output.In an embodiment, the sketch construction module 110 enables fine-tunedmanipulation of facial components through interactive bounding boxes, providingpixel-level alignment accuracy.The image enhancement module 112 is operatively coupled to the sketchconstruction module 110 and configured to receive the digital sketch image fromthe sketch construction module 110.In an embodiment, the image enhancement module 112 comprises a deep learningmodel implemented by a PyTorch-based Generative Adversarial Network (GAN),the GAN comprising a U-Net generator and a PatchGAN discriminator trained onthe Chinese University of Hong Kong (CUHK) and Indian Face Sketch datasets,and an image processing engine implementing an OpenGV configured to applycolor corrections, brightness, adjustment, and sharpening operations, wherein theenhancement module 112 outputs a photorealistic face image. The OpenCV engineapplies post-processing enhancements including histogram equalization, Gaussiansharpening, and CLAHE (Contrast Limited Adaptive Histogram Equalization).In an embodiment, the image enhancement module 112 is trained using pairedimage datasets, including CUHK Face Sketch Dataset (CUFS) and the Indian FaceSketch Database.The image storage and cloud indexing module 114 is operatively connected to theimage enhancement module 112 and configured to store the photorealistic faceimage in a cloud-based storage service, dynamically index the stored image into thecloud-based storage service for face recognition purposes and provide the indexedimage as an input for recognition.In an embodiment, the image storage and cloud indexing module 114 maintainsversion control of enhanced images using unique file hashes and supportsasynchronous indexing into the cloud-based storage service.The recognition module 116 is functionally integrated with the image storage andcloud indexing module 114, and comprises the GUI 110a built using Java Swingfor user interaction, configured to receive as input an image selected by the user,wherein the input image includes a freshly created sketch, a GAN-enhanced image,or an external hand-drawn sketch, perform a face matching operation using thecloud-based storage service to compare the input image against indexed faces andcompute similarity scores, display the matched image with similarity score if thesimilarity score is above a predefined threshold, and otherwise display a "No MatchFound".In an embodiment, the recognition module 116 is configured to return a ranked listof top-N matches, where N is user-defined, each with a similarity score andassociated metadata. The average response time of the recognition module 116 forprocessing and displaying results for input images under 500 KB in size is less than2 seconds.In an embodiment, the system enables real-time sketch creation, enhancement,cloud-based storage, and face recognition to assist in forensic investigations.In an embodiment, the system 100 for forensic face sketch generation,enhancement, and recognition comprises the user interface (UI) module 102operable on the remote device 104. The UI 102 includes the user authenticationsub-module 106 supporting two secure login modes, one using OTP via email andthe other using SHA-256-hashed credentials. Credentials are stored in a MySQLdatabase using BCrypt and SHA-256. The sketch construction module 110, builtwith JavaFX, allows users to create composite sketches by dragging and droppingfacial components such as eyes, noses, and lips, offering pixel-level manipulationand external sketch integration. The image enhancement module 112 uses aPyTorch-based GAN model (U-Net generator and PatchGAN discriminator)trained on CUHK and Indian face sketch datasets, combined with OpenCV-basedprocessing for color correction, sharpening, and histogram equalization to producephotorealistic images. Enhanced images are stored and indexed in cloud storage viathe image storage and cloud indexing module 114, which ensures version controlthrough file hashes. The recognition module 116, with a Java Swing GUI, matchesuser-provided sketches or enhanced images with cloud-stored data, returningsimilarity scores and ranked matches within two seconds. The integrated system100 supports real-time sketch creation, enhancement, and recognition, aiding inefficient forensic investigations.The disclosed system 100 further comprises hardware components including one ormore processors, memory units, and data storage modules, configured to executethe remote device 104, the user interface (UI) module 102, the user authenticationsub-module 106, the sketch construction module 110, the image enhancementmodule 112, the image storage and cloud indexing module 114, and the recognitionmodule 116.In an embodiment, these modules work collaboratively to enable secure login,interactive sketch creation, deep learning-based image enhancement, cloud-basedimage indexing, and high-speed facial recognition. The system 100 architectureensures seamless coordination between modules, real-time data processing, andoptimized performance for forensic face identification workflows.In an embodiment, the system 100 can include one or more processors and may beimplemented as one or more microprocessors, microcomputers, microcontrollers,digital signal processors, central processing units, logic circuitries, and / or anydevices that manipulate data based on operational instructions. Among othercapabilities, the one or more processors are configured to fetch and executecomputer-readable instructions stored in a memory of the system 100. The memorymay store one or more computer-readable instructions or routines, which may befetched and executed for executing the instructions. The memory may include anynon-transitory storage device including, for example, volatile memory such asRAM, or non-volatile memory such as EPROM, flash memory, and the like. Thefunctions of one or more processor(s) may be provided through the use of dedicatedhardware as well as hardware capable of executing machine-readable instructions.In other examples, one or more processors may be implemented by electroniccircuitry or printed circuit board. One or more processors may be configured toexecute functions of various modules of the system 100 such as the remote device104, the user interface (UI) module 102, the user authentication sub-module 106,the sketch construction module 110, the image enhancement module 112, the imagestorage and cloud indexing module 114, and the recognition module 116.In an alternative aspect, the memory may be an external data storage device coupledto the system 100 directly or through one or more offline / online data servers.In an embodiment, the system 100 further comprises a network interface to receivereal-time data inputs from external sources such as databases, APIs, and sensors,which are used by the image enhancement module 112 and the recognition module116 to dynamically update system parameters and improve face sketchenhancement and recognition accuracy. The network interface may include avariety of interfaces, for example, interfaces for data input and output devices,referred to as I / O devices, transceivers, storage devices, and the like. The networkinterface may facilitate communication of the system 100 with various devicescoupled to the system 100. The network interface may also provide acommunication pathway for one or more components of the system 100. Examplesof such components include, but are not limited to, processing module(s) and datastorage.The processing modules(s) may be implemented as a combination of hardware andprogramming (for example, programmable instructions) to implement one or morefunctionalities of the processing module(s). In the examples described herein, suchcombinations of hardware and programming may be implemented in severaldifferent ways. For example, the programming for the processing module(s) maybe processor-executable instructions stored on a non-transitory machine-readablestorage medium and the hardware for the processing module(s) may include aprocessing resource (for example, one or more processors), to execute suchinstructions. In the present examples, the machine-readable storage medium maystore instructions that, when executed by the processing resource, implement theprocessing module(s). In such examples, the system 100 may include the machinereadable storage medium storing the instructions and the processing resource toexecute the instructions, or the machine-readable storage medium may be separatebut accessible to the system 100 and the processing resource. In other examples, theprocessing module(s) may be implemented by electronic circuitry and the remotedevice 104, the user interface (UI) module 102, the user authentication sub-module106, the sketch construction module 110, the image enhancement module 112, theimage storage and cloud indexing module 114, and the recognition module 116.In an embodiment, the system 100 can be deployed to law enforcement officers,crime scene investigators, Intelligence agencies, private security firms, and defenseorganizations. Additionally, it can be adapted for use in border control,immigration, or airport security agencies. Further, our developed system can serveeducational institutions for research and training in forensic technology as abenchmark to improve upon.Figures 2A and 2B illustrate a flow chart depicting the steps involved in method200 for end-to-end forensic face identification through a composite sketch image inaccordance with an embodiment of the present disclosure. The order in whichmethod 200 is described is not intended to be construed as a limitation, and anynumber of the described method steps may be combined in any order to implementmethod 200, or an alternative method. Furthermore, method 200 may beimplemented by processing resource or computing device(s) through any suitablehardware, or a combination thereof. The method 200 comprises the following steps:At step 202, the method 200 includes authenticating, by a user interface module102, a user via credential hashing and OTP validation.At step 204, the method 200 includes constructing, by a sketch constructionmodule 110, a suspect facial sketch using a drag-and-drop GUI that enablesselection and manipulation of facial components.At step 206, the method 200 includes enhancing, by an image enhancementmodule 112, the constructed sketch using a Generative Adversarial Network(GAN) model, wherein the GAN includes a U-Net-based generator and aPatchGAN-based discriminator.At step 208, the method 200 includes storing, by an image storage and cloudindexing module 114, the enhanced image in a cloud-based storage service.At step 210, the method 200 includes indexing, by an image storage and cloudindexing module 114, the stored image into the cloud-based storage service.At step 212, the method 200 includes comparing, by a recognition module 116,an uploaded image with images in the cloud-based storage service usingsimilarity thresholds.At step 214, the method 200 includes displaying on a GUI 110a the matchingresult with a similarity percentage if a match is found, or displaying a "NoMatch Found" message otherwise.Figure 3 illustrates the operational flow of the system for forensic face sketchgeneration, enhancement, and recognition in accordance with the presentdisclosure. The process begins with a secure login, where the user interface (UI)module 102 ensures that only authorized personnel, such as law enforcementofficers or forensic investigators, can access the system 100. Following this, thesystem 100 offers two options for rough sketch input or selection of facial featuresthrough the sketch construction module 110. If an eyewitness sketch is available, itcan be uploaded directly to the sketch construction module 110 otherwise, the usercan create a facial composite by selecting predefined facial features (eyes, nose,mouth, hairstyle, etc.) through the graphical user interface (GUI) 110a. After thesketch is created or uploaded, the system 100 generates an obtained sketch and arendered image.The image enhancement module 112 processes this sketch using generative or deeplearning-based algorithms to enhance its visual clarity and create a photo-realisticrepresentation. The enhanced image is then exported to the feature extraction,where facial encoding and vector embedding techniques extract unique biometrictraits, preparing it for database comparison. These extracted features are thencompared against a centralized or cloud-based criminal database using advancedface recognition algorithms, performing similarity matching to identify potentialmatches based on facial traits or sketch similarity.Finally, the system 100 displays the recognition result on the GUI 110a, showingpotential suspects or matched individuals along with confidence scores, helpinginvestigators narrow down search results efficiently, even when starting from vagueor non-photographic information. This operational flow ensures seamlessintegration of all core functionalities, secure access, sketch creation, imageenhancement, feature extraction, database comparison, and result display, makingthe system 100 an invaluable tool for criminal investigations, border security, andmissing person's identification.Figure 4 illustrates the sketch construction flow in accordance with the presentdisclosure. The sketch construction module 110 and the image enhancementmodule 112 are integrated into a sketch creation. The process commences with theauthorized user initiating a new facial sketch creation session and validatingthrough the sketch construction login interface 106a, wherein the user selectsindividual facial elements such as eyes, nose, mouth, or hairstyle from a predefinedrepository. Each selected element is further refined by choosing specific featuretypes aligned with the witness's description. A decision gateway evaluates whetherthe chosen features accurately reflect the eyewitness account. In cases of mismatch,the user is prompted to reselect alternative options, maintaining the integrity of thecomposite under the principle of accurate representation. This iterative selectioncontinues until all facial components are defined and validated. Upon confirmingcompletion, the final sketch is saved and becomes a system-generated evidentiaryasset, eligible for further biometric processing and legal scrutiny. The process notonly supports modular construction and correction but also upholds forensicaccountability, making it admissible in legal proceedings and suitable for lawenforcement investigations.Figure 5 illustrates the recognition flow in accordance with the present disclosure.This process begins with the authorized user initiating the recognition module 116functionally integrated with the image storage and cloud indexing module 114, andcomprising the GUI 110a built using Java Swing for user interaction, ensuring theoperation is performed within a secured and auditable environment, therebysatisfying chain-of-custody and data integrity requirements.The recognition module 116 then prompts for the sketch type, allowing input eitherfrom a manually constructed composite or an eyewitness-uploaded facial sketch.Upon determining the input type, the sketch is routed through a digital processinginterface where it undergoes pre-processing, encoding, and biometric featureextraction. These features are then cross-referenced against a secure criminaldatabase using advanced facial recognition algorithms that comply with dataprotection and law enforcement standards.Following the matching phase, the GUI 110a computers display similarity scores,offering quantifiable evidence of resemblance between the input sketch and existingrecords. When a potential match is identified, the GUI 110a proceeds to displaysuspect details, including metadata such as name, previous offenses, andidentification numbers, where available. This information, coupled with theconfidence score, supports investigative and prosecutorial decision-making. Thisrecognition flow ensures traceability, accuracy, and lawful processing, making itsuitable for criminal investigations, surveillance, border security, and missingpersons identification under strict judicial and ethical frameworks.Figures 6A-6F illustrate the user interface (UI) sketch construction of the two-stepverification, sketch construction canvas page, drag and drop features, sketch to facegeneration (enhancement), and recognition scenes in accordance with the presentdisclosure. Figure 6A illustrates the two-step verification process for the sketchconstruction module 110, where users undergo a dual-layered authenticationinvolving BCrypt-hashed password validation followed by a One-Time Password(OTP) delivered via the Jakarta Mail API, ensuring secure access to the sensitivesketching interface.Figure 6B presents the sketch construction modules 110 canvas page, a JavaFXbased GUI 110a where users can interactively drag and drop facial componentssuch as head shapes, eyes, noses, lips, and eyebrows onto a virtual canvas, witheach feature adjustable via resizable and movable bounding boxes for precise sketchcreation.Figure 6C highlights the drag-and-drop functionality, emphasizing the intuitiveplacement and manipulation of facial features, which allows non-artist users toconstruct detailed suspect sketches rapidly based on eyewitness descriptions.Figure 6D demonstrates the sketch-to-face generation process within the imageenhancement module 112, where a U-Net-based Generative Adversarial Network(GAN) paired with a PatchGAN discriminator transforms a rough sketch into aphotorealistic face image, further refined through OpenCV post-processingtechniques like color correction and sharpening to enhance visual realism.Figures 6E and 6F depict the recognition module's 116 interface and outcomes,showcasing the Java Swing-based GUI 110a where AWS Rekognition compares anuploaded sketch or enhanced image against a cloud-stored face database in an AWSS3 bucket, displaying matched images with similarity percentages when the matchexceeds a predefined threshold (e.g., 70%), or indicating "No Match Found"otherwise. Collectively, these figures underscore the system's 100 modulararchitecture, user-friendly design, and high forensic precision, enabling efficientsuspect identification through a seamless integration of advanced user interfaces,deep learning, and cloud-based recognition technologies.Figure 7 illustrates the recognition scene of the hand-drawn sketch in accordancewith the present disclosure. The recognition scenario within the recognition module116 effectively matches a hand-drawn sketch to a corresponding facial image froma database. The GUI 110a comprises three key sections: the left panel displays theinput sketch based on eyewitness description, the center panel shows the matchedfacial image with a calculated similarity score of 75.90%, and the right panelpresents detailed match results, including the file name, confidence score, andpositional coordinates of the detected face. Supporting functionalities such as"Open Sketch," "Upload to S3," "Find Match," and "Index Faces" enablestreamlined user interaction, secure data handling, and efficient databasemanagement. This recognition module 116 exemplifies the integration of traditionalforensic sketching with advanced AI-driven face recognition technology, allowinglaw enforcement and forensic professionals to identify suspects more accuratelyand legally validate evidence through a traceable, secure, and technologicallyrobust interface.In an operative configuration, system 100 comprises the remote computing device104 and begins with the user interface (UI) module 102, which includes the userauthentication sub-module 106. This sub-module supports two modes of secureaccess: in the first mode, login credentials are verified against hashed values in thedatabase 108, and an OTP is sent via email API to the user's registered email forverification through the sketch construction login interface 106a; in the secondmode, credentials are validated using SHA-256 hashing for direct access throughthe recognition login interface 106b. Once authenticated, users can access thesketch construction module 110, which provides the JavaFX-based GUI 110a tobuild composite face sketches using a drag-and-drop library of facial componentssuch as eyes, noses, lips, head shapes, eyebrows, and identification marks. Themodule supports resizing, rotation, repositioning, and integration of external handdrawn sketches, resulting in a digitally constructed sketch image. This image ispassed to the image enhancement module 112, which uses a PyTorch-based GANwith a U-Net generator and PatchGAN discriminator trained on CUHK and IndianFace Sketch datasets to generate a photorealistic face. Additionally, an OpenGVpowered image engine enhances color, brightness, and sharpness. The enhancedimage is stored and indexed by the image storage and cloud indexing module 114in a cloud-based system, preparing it for recognition. The recognition module 116,built with a Java Swing GUI, allows users to upload a sketch, GAN-enhancedimage, or external drawing for comparison against cloud-stored faces. It calculatessimilarity scores and displays a matched face if the score exceeds a predefinedthreshold, otherwise, it shows "No Match Found." Overall, System 100 enablesreal-time, secure, and efficient creation, enhancement, storage, and recognition offorensic face sketches, greatly enhancing the accuracy and speed of criminalinvestigations.Advantageously, the system 100 offers a solution for forensic investigations byintegrating sketch creation, image enhancement, and face recognition into a singleplatform. It features a user-friendly interface, robust authentication, and securecloud-based access, enabling remote usage across crime units. The systemautomates the process from composite sketching to suspect identification,significantly reducing investigation time. Its flexibility allows easy deployment andexpansion. Enhanced by deep learning, it converts rough or hand-drawn sketchesinto realistic, photo-like images, improving match accuracy with suspect databases.This results in faster, more reliable identifications, making the system a valuabletool for modern law enforcement agencies.The functions described herein may be implemented in hardware, softwareexecuted by a processor, firmware, or any combination thereof. If implemented insoftware executed by a processor, the functions may be stored on or transmittedover as one or more instructions or codes on a computer-readable medium. Otherexamples and implementations are within the scope and spirit of the disclosure andappended claims. For example, due to the nature of software, functions describedabove can be implemented using software executed by a processor, hardware,firmware, hardwiring, or combinations of any of these. Features implementingfunctions may also be physically located at various positions, including beingdistributed such that portions of functions are implemented at different physicallocations.The foregoing description of the embodiments has been provided for purposes ofillustration and is not intended to limit the scope of the present disclosure.Individual components of a particular embodiment are generally not limited to thatparticular embodiment but are interchangeable. Such variations are not to beregarded as a departure from the present disclosure, and all such modifications areconsidered to be within the scope of the present disclosure.TECHNICAL ADVANCEMENTSThe present disclosure described herein above has several technical advantagesincluding, but not limited to, the realization of a system for forensic face sketchgeneration, enhancement, and recognition, and a method thereof that:- enables end-to-end forensic processing by integrating sketch creation,image enhancement, and face recognition into a single unified platform;- has a user-friendly interface;- offers cloud scalability and reliability;- reduces investigation time by automating the entire process from sketchcreation to suspect matching;- allows straightforward expansion, data security, and remote access fromdifferent crime units anywhere;- offers robust authentication and data protection;- is flexible and easy to deploy;- increases recognition success by generating high-quality images that bettermatch real-world suspect databases;- significantly reduces investigation time;- creates composite facial sketches based on verbal eyewitness accounts; and- enhances rough sketches into realistic, photo-like images for better visualclarity.The embodiments herein and the various features and advantageous details thereofare explained with reference to the non-limiting embodiments in the followingdescription. Descriptions of well-known components and processing techniques areomitted so as to not unnecessarily obscure the embodiments herein. The examplesused herein are intended merely to facilitate an understanding of ways in which theembodiments herein may be practiced and to further enable those of skill in the artto practice the embodiments herein. Accordingly, the examples should not beconstrued as limiting the scope of the embodiments herein.The foregoing description of the specific embodiments so fully reveals the generalnature of the embodiments herein that others can, by applying current knowledge,readily modify and / or adapt for various applications such specific embodimentswithout departing from the generic concept, and, therefore, such adaptations andmodifications should and are intended to be comprehended within the meaning andrange of equivalents of the disclosed embodiments. It is to be understood that thephraseology or terminology employed herein is for the purpose of description andnot of limitation. Therefore, while the embodiments herein have been described interms of preferred embodiments, those skilled in the art will recognize that theembodiments herein can be practiced with modification within the spirit and scopeof the embodiments as described herein.The use of the expression "at least" or "at least one" suggests the use of one or moreelements or ingredients or quantities, as the use may be in the embodiment of thedisclosure to achieve one or more of the desired objects or results.While considerable emphasis has been placed herein on the components andcomponent parts of the preferred embodiments, it will be appreciated that manyembodiments can be made and that many changes can be made in the preferredembodiments without departing from the principles of the disclosure. These andother changes in the preferred embodiment as well as other embodiments of thedisclosure, will be apparent to those skilled in the art from the disclosure herein,whereby it is to be distinctly understood that the foregoing descriptive matter is tobe interpreted merely as illustrative of the disclosure and not as a limitation.

Claims

1. A system for forensic face sketch generation, enhancement, and recognition, said system (100) comprising: - a user interface (UI) module (102) operable on a remote device (104), wherein said UI module (102) comprising: o a user authentication sub-module (106) configured to: - receive login credentials from an authorized user, - verify the credentials against hashed values stored in a database (108), - in a first mode, dispatch an on-time password (OTP) via an email application program interface (API) to a registered email of the user for verification through a sketch construction login interface (106a), and - in a second mode, validate credentials hashed with SHA-256 (Secure Hash Algorithm 256-bit) for access via a recognition login interface (106b); - a sketch construction module (110) operatively connected to said user authentication sub-module (106) and comprising a graphical user interface (GUI) (110a) built using JavaFX, configured to: - allow said user to construct a composite face sketch by selecting, dragging, and dropping facial components selected from a predefined component set, including at least eyes, noses, lips, head shapes, eyebrows, and identification marks, - provide resizing, repositioning, and rotation functionalities using bounding box controls, - optionally accept external hand-drawn sketch images for integration, and - generate a digital sketch image as output; - an image enhancement module (112) operatively coupled to said sketch construction module (110) and configured to receive the digital sketch image from the sketch construction module (110), wherein the image enhancement module (112) comprising: - a deep learning model implemented by a PyTorch-based Generative Adversarial Network (GAN), said GAN comprising a U-Net generator and a PatchGAN discriminator trained on the Chinese University of Hong Kong (CUHK) and Indian Face Sketch datasets, and - an image processing engine implementing an OpenGV configured to apply color corrections, brightness, adjustment, and sharpening operations, wherein said enhancement module (112) outputs a photorealistic face image; - an image storage and cloud indexing module (114) operatively connected to said image enhancement module (112), and configured to: - store said photorealistic face image in a cloud-based storage service, - dynamically index the stored image into the cloud-based storage service for face recognition purposes, and - provide the indexed image as an input for recognition; and - a recognition module (116) functionally integrated with said image storage and cloud indexing module (114), and comprising a GUI built using Java Swing for user interaction, configured to: - receive as input an image selected by said user, wherein said input image includes a freshly created sketch, a GANenhanced image, or an external hand-drawn sketch, - perform a face matching operation using the cloud-based storage service to compare said input image against indexed faces and compute similarity scores, - display the matched image with similarity score if the similarity score is above a predefined threshold, and otherwise display a "No Match Found"; - wherein said system (100) enables real-time sketch creation, enhancement, cloud-based storage, and face recognition to assist in forensic investigations.

2. The system (100) as claimed in claim 1, wherein said database (108) is a MySQL database used to store credentials for both sketch construction and recognition modules, hashed using BCrypt and SHA-256, respectively.

3. The system (100) as claimed in claim 1, wherein the sketch construction module (110) enables fine-tuned manipulation of facial components through interactive bounding boxes providing pixel-level alignment accuracy..

4. The system (100) as claimed in claim 1, wherein the OTP in said sketch module login interface (106a) is transmitted using Jakarta Mail API, forming a two-factor authentication mechanism integrated with hashed password validation.

5. The system (100) as claimed in claim 1, wherein the image enhancement module (112) is trained using paired image datasets, including CUHK Face Sketch Dataset (CUFS) and the Indian Face Sketch Database.

6. The system (100) as claimed in claim 1, wherein the OpenCV engine applies post-processing enhancements including histogram equalization, Gaussian sharpening, and CLAHE (Contrast Limited Adaptive Histogram Equalization).

7. The system (100) as claimed in claim 1, wherein the image storage and cloud indexing module (114) maintains version control of enhanced images using unique file hashes, and supports asynchronous indexing into the cloud-based storage service.

8. The system (100) as claimed in claim 1, wherein the recognition module (116) is configured to return a ranked list of top-N matches, where N is userdefined, each with a similarity score and associated metadata9. The system (100) as claimed in claim 1, wherein the average response time of the recognition module (116) for processing and displaying results for input images under 500 KB in size is less than 2 seconds.

10. A method (200) for end-to-end forensic face identification through a composite sketch image, said method (200) comprising: - authenticating, by a user interface module (102), a user via credential hashing and OTP validation; - constructing, by a sketch construction module (110), a suspect facial sketch using a drag-and-drop GUI that enables selection and manipulation of facial components; - enhancing, by an image enhancement module (112), the constructed sketch using a Generative Adversarial Network (GAN) model, wherein the GAN includes a U-Net-based generator and a PatchGAN-based discriminator; - storing, by an image storage and cloud indexing module (114), the enhanced image in a cloud-based storage service; - indexing, by an image storage and cloud indexing module (114), the stored image into the cloud-based storage service; - comparing, by a recognition module (116), an uploaded image with images in the cloud-based storage service using similarity thresholds; and - displaying on a GUI the matching result with a similarity percentage if a match is found, or displaying a "No Match Found" message otherwise.