Volume 9, Issue 4, April – 2024 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24APR1534

Handwritten Calculator Using

Contour Detection
Dr. S. Venkata Ramana1; T. Sai Ram Charan2; S. Shanmukha Pavan3; R. Czar Raju4; T. Dileep Kumar5
12345
Sagi Rama Krishnam Raju Engineering College (A), Bhimavaram, Andhra Pradesh, India

Abstract:- This project introduces a novel Handwritten II. LITERATURE SURVEY

Calculator that merges traditional handwritten
calculations with modern computing techniques. By The integration of Optical Character Recognition (OCR)
leveraging computer vision, machine learning, and technology into mathematical computation has been explored
human-computer interaction, the calculator interprets in several studies, each presenting unique perspectives and
handwritten digits and operators to perform approaches. Supriadi S's paper introduced a Handwriting
mathematical calculations. Through contour detection and Calculator application leveraging OCR and Contour Detection
Convolutional Neural Network (CNN) model prediction, algorithms, aiming to streamline mathematical calculations
the calculator offers users a seamless and intuitive way to through intuitive input methods [11]. Similarly, Muhammad
compute. The project explores the development, Aamir's study emphasized the efficiency and convenience of
applications, and impact of this innovative technology, OCR-based Handwriting Calculator applications, highlighting
with potential implications for education, accessibility, their transformative potential in computational tasks [12].
and software engineering. The Handwritten Calculator Furthermore, Sajjad Hussain et al.'s research showcased the
represents a significant advancement in computational application of Convolutional Neural Networks (CNNs) in
methods, bridging the gap between analog and digital handwritten digit recognition, underscoring the advancements
approaches to calculation. facilitated by deep learning techniques in pattern recognition
technology [3]. Together, these studies reflect a trend towards
Keywords:- Contour Detection, CNN, Image Processing etc. leveraging OCR and advanced algorithms to enhance
mathematical computation and digit recognition tasks.
I. INTRODUCTION
Venkataramana S's study underscores the significance of
The evolution of computational tools, from Blaise making mathematics more accessible and intuitive through
Pascal's invention of the calculator in 1642 to modern-day OCR technology, offering a seamless alternative to traditional
sophisticated devices, has revolutionized various domains, calculators [7]. Muhammad Aamir's research further
including business transactions [1,2]. However, handwritten emphasizes the transformative impact of OCR-based
calculations present challenges, necessitating the integration Handwriting Calculator applications in streamlining
of Optical Character Recognition (OCR) technology to mathematical computations, redefining user engagement with
streamline mathematical tasks. In response, recent projects numerical tasks [2]. Additionally, Sajjad Hussain et al.'s paper
aim to segment and recognize handwritten mathematical highlights the groundbreaking advancements facilitated by
expressions, leveraging Convolutional Neural Network CNNs in handwritten digit recognition, paving the way for
(CNN) models to ensure deployability across diverse automation and efficiency in pattern recognition tasks [3].
computational devices [3,5]. The convergence of OCR Collectively, these studies demonstrate the potential of OCR
technology and mathematical computation promises enhanced technology and advanced algorithms in revolutionizing
efficiency and accessibility, driving advancements towards a computational tasks and driving advancements in pattern
paperless society. recognition technology.

As society moves towards digitization, the need for III. SYSTEM IMPLEMENTATION
accurate handwritten data recognition interfaces becomes
increasingly vital [4,6]. Handwriting recognition, both online A. System Overview:
and offline, is essential for effectively digitizing documents The proposed system aims to develop a handwriting
and enabling efficient data retrieval and manipulation. calculator application that utilizes contour detection
However, multilingual contexts and diverse writing systems algorithms to recognize handwritten mathematical expressions
present unique challenges, necessitating ongoing research and evaluate them. This system will allow users to perform
efforts to develop robust OCR solutions capable of mathematical calculations simply by writing expressions on
accommodating linguistic complexities [2, 5]. By addressing the user interface. The application will then capture the
these challenges and integrating OCR technology into handwritten input, process it using contour detection
mathematical computation frameworks, researchers are not technique, and translate it into digital format for computation.
only enhancing computational efficiency but also empowering
users to seamlessly convert handwritten expressions into
digital format for streamlined mathematical tasks [8,9,10].

IJISRT24APR1534 www.ijisrt.com 1307

Volume 9, Issue 4, April – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24APR1534

B. Architecture:  Contour Detection Module:

This module will employ contour detection algorithms to
identify and extract the handwritten characters from the pre-
processed input image. Edge detection technique in Contour
detection, will be applied to locate the boundaries of
individual characters.

 Character Recognition Module:

Once the contours of handwritten characters are
detected, this module will perform character recognition to
identify the corresponding mathematical symbols or digits. A
Convolutional Neural Network (CNN) model, trained on a
dataset of handwritten characters, is employed for accurate
recognition.

 Expression Parsing Module:

After recognizing individual characters, this module will
parse the detected symbols to reconstruct the mathematical
expression. It will handle the arrangement and interpretation
of characters based on mathematical rules to ensure the
correct representation of the input expression.

 Computation Module:
Once the mathematical expression is parsed, this module
will evaluate the expression to compute the result. It will
utilize mathematical evaluation algorithms to perform
arithmetic operations based on the parsed expression.

 Output Module:
Finally, the computed result will be presented to the user
through the user interface. Users are provided with the
solution to their mathematical expression.

IV. RESULTS & DISCUSSIONS

A. Results:
The system implementation successfully yielded a fully
functional Handwritten Calculator application. The
application is capable of:

 Accepting handwritten mathematical expressions as input.

 Preprocessing the captured image to enhance character
clarity for improved recognition.
 Employing the trained machine learning model to identify
individual characters within the expression.
 Parsing the recognized characters and symbols,
Fig 1 Architecture interpreting the order of operations to build a symbolic
representation.
 Input Module:  Executing calculations based on the parsed expression.
This module is responsible for capturing handwritten  Presenting the final calculated result on the user interface.
input from the user interface. Input can be received through
various devices such as touchscreens, digital pens, or cursor
control.

 Preprocessing Module:
Before contour detection can be applied, the handwritten
input needs to be pre-processed to enhance its quality and
prepare it for analysis. Preprocessing techniques include
image resizing, noise reduction, and binarization to ensure Fig 2 Contours of Each Digit and Operator, Along with the
optimal performance of the contour detection algorithm. Accuracy of our Model's Predictions for Each

IJISRT24APR1534 www.ijisrt.com 1308

Volume 9, Issue 4, April – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24APR1534

Fig 3 Sample Output – 1 Fig 5 Sample Output – 3

Fig 4 Sample Output – 2

B. Discussion:

Table 1 Comparison of Proposed Model with Existing Models

MODEL CHARACTERISTICS
Simple Handwriting Calculator Character Recognition: Digits (0-9) and arithmetic symbols (+, -, *, /)
Application Using Optical Expression Evaluation: Evaluates expressions with digits and basic arithmetic symbols
Character Recognition (OCR) – Strengths: Simple and easy to implement
Supriadi Limitations: Limited character set and expression complexity
Towards Handwritten Character Recognition: Mathematical expressions Expression Evaluation: Evaluation not
Mathematical Expression mentioned (focuses on recognition) Strengths: Potentially broader recognition capabilities
Recognition | IEEE Xplore Limitations: Evaluation capabilities unclear
Character Recognition: Digits (0-9), basic arithmetic symbols (+, -, *, /), and brackets ( )
Proposed Model Expression Evaluation: Evaluates expressions with digits, arithmetic symbols, and brackets
Strengths: Recognizes a wider range of characters, including brackets, for more complex
expressions Limitations: May require a more complex model for accurate bracket
recognition and handling

V. CONCLUSION  The Application Offers Several Advantages Over

Traditional Calculators:
This project successfully developed a functional
Handwritten Calculator application utilizing contour detection  Increased User Interaction: The ability to use
and machine learning for character recognition. The handwritten input provides a more natural and user-
application offers an intuitive user experience by allowing friendly experience.
users to input mathematical expressions through handwritten  Reduced Errors: Eliminates the need for manual data
images. It demonstrates the feasibility of this approach for entry, minimizing typos or errors during input.
real-time processing and evaluation of mathematical  Improved Accessibility: Caters to users with physical
expressions. limitations who might struggle with traditional
calculators.

IJISRT24APR1534 www.ijisrt.com 1309

Volume 9, Issue 4, April – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24APR1534

 Potential for Educational Applications: Provides an [5]. B Uma Maheswari, S. V. Chiranjeevi, C. Sushama, S.
interactive tool for practicing math concepts and problem- Venkataramana, & D Naga Malleswari. (2022).
solving skills. Malaria cell detection using deep learning techniques
 Exploration of Complex Expressions: The recognition and Investigation on efficacy and safety of carcia
of brackets allows for evaluation of more intricate papaya leaf extract on malaria. Journal of
mathematical expressions. Pharmaceutical Negative Results, 50–57.
[6]. Yang, Jimei & Price, Brian & Cohen, Scott & Lee,
FUTURE WORK Honglak & Yang, Ming-Hsuan. (2016). Object
Contour Detection with a Fully Convolutional
 While the Application Demonstrates Promising Results, Encoder-Decoder Network. 193-202.
there's Room for Further Exploration and Improvement: 10.1109/CVPR.2016.28.
[7]. S, Venkataramana. (2024). Recognition of Human
 Accuracy Enhancement: Being through Handwritten Digits Using Image
Explore more sophisticated machine learning models or Processing Techniques and AI. 9. 651-654.
training on a larger and more diverse dataset to improve 10.48047/IJIEMR/V09/I12/109.
recognition accuracy. [8]. Petrou, Maria & Kovalev, Vassili & Reichenbach,
Jürgen. (2006). Three-Dimensional Nonlinear
 Robustness Improvement: Invisible Boundary Detection. IEEE transactions on
Investigate techniques to enhance the application's image processing : a publication of the IEEE Signal
performance in handling variations in handwriting styles, Processing Society. 15. 3020-32.
complex symbols. 10.1109/TIP.2006.877516.
[9]. Jain, Mohit & Mathew, Minesh & Jawahar, C.V..
 Real-time Optimization: (2017). Unconstrained OCR for Urdu Using Deep
For real-time applications, optimize the code to ensure CNN-RNN Hybrid Networks. 747-752.
minimal processing delays. 10.1109/ACPR.2017.5.
[10]. MISHRA, Dr. (2024). NEURAL NETWORKS AND
 Symbol Recognition Expansion: DEEP LEARNING: THEORITICAL INSIGHTS
Extend the model's capabilities to recognize a wider AND FRAMEWORKS.
range of mathematical symbols. 10.61909/AMKEDTB022409.
[11]. Desai, N & S, Venkataramana & Sekhar, B V D S.
 Feedback Mechanism Development: (2020). Automatic Visual Sentiment Analysis with
Convolution Neural Network. 31. 351-360.
Implement a comprehensive feedback mechanism to
guide users in case of recognition errors or invalid 10.22068/ijiepr.31.3.351.
[12]. Text entry systems: mobility, accessibility,
expressions.
universality. San Francisco, Calif: Morgan Kaufmann,
By addressing these areas, the Handwritten Calculator 2007. doi: 10.1016/B978-0-12-373591-1.X5000-1.
application can evolve into a robust, user-friendly, and
versatile tool for mathematical problem-solving and learning.

REFERENCES

[1]. Supriadi, Supriadi. (2021). Simple Handwriting

Calculator Application Using Optical Character
Recognition (OCR). Buletin Ilmiah Sarjana Teknik
Elektro. 2. 163. 10.12928/biste.v2i3.3348.
[2]. Memon, Jamshed & Sami, Maira & Khan, Rizwan &
Uddin, Mueen. (2020). Handwritten Optical Character
Recognition (OCR): A Comprehensive Systematic
Literature Review (SLR). IEEE Access. 1-1.
10.1109/ACCESS.2020.3012542.
[3]. Jain, Mayank & Kaur, Gagandeep & Quamar,
Muhammad & Gupta, Harshit. (2021). Handwritten
Digit Recognition Using CNN. 211-215.
10.1109/ICIPTM52218.2021.9388351.
[4]. Tran, Giang & Huynh, Kien & Le, Thanh-Sach &
Phan, Tan-Phuc & Bui, Khanh-Ngoc. (2018).
Handwritten Mathematical Expression Recognition
Using Convolutional Neural Network. 15-19.
10.1109/CRC.2018.00012.

IJISRT24APR1534 www.ijisrt.com 1310