Ankur Choudhary
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Ankur Choudhary has a total of 46 co-authors. Here are the top co-authors by number of shared publications:
Anurag Singh Baghel
Department of Computer Science and Engineering School of ICT, Gautam Buddha University Uttar Pradesh India
6 shared publicationsOm Prakash Sangwan
Department of Computer Science and Engineering Guru Jambheshwar University Haryana India
4 shared publications- Arun Prakash Agrawal
School of Computer Science Engineering and Technology Bennett University Greater Noida India
2 shared publications - Arvind Kumar
Bennett University
2 shared publications - Parma Nand
Sharda University
2 shared publications Pragya Agarwal
CS &E Department, Amity University, Pitney Bowes Software, CS &E Department, Amity University, sec-125/44, Noida, U.P., India
2 shared publications
Latest Publications
- Open Access
- 30 September 2026
- Published by Taru Publications in Journal of Interdisciplinary Mathematics
- Vol. 29 (9) , 2807-2820
A mathematical framework for medical image security using DWT-based LL sub-band coefficient analysis and robust watermark embedding
With the advancement of telemedicine, it is very important to preserve the authenticity of the medical image, shared remotely. Watermarking is commonly used to...
- 30 September 2026
- Published by Taru Publications in Journal of Interdisciplinary Mathematics
- Vol. 29 (9) , 2807-2820
Open AccessA mathematical framework for medical image security using DWT-based LL sub-band coefficient analysis and robust watermark embedding
With the advancement of telemedicine, it is very important to preserve the authenticity of the medical image, shared remotely. Watermarking is commonly used to...
- Open Access
- 1 December 2025
- Published by The ScienceIn Publishing in Journal of Integrated Science and Technology
- Vol. 14 (1)
Secure biometric watermarking in medical images using LH Sub-band Pixel Pair Mapping
Securing a medical image is important and crucial task over digital platform. In this research work, we have proposed a novel method to embed the biometric watermark into medical image. In this process, medical images are more secure for sharing information with medical practitioners over internet. First, we have performed rotation and flipping of medical image. after that Utilization of Discrete Wavelet Transform(DWT) with Haar wavelets decomposes the images into four sub-bands. Watermark is securely embedded in the LH sub band. Now, we have applied Pixel Pair Mapping (PPM) in LH to conceal information bits. Without inverse, the original image can not be retrieved. Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index Measure (SSIM), and Normalized Cross-Correlation (NC) have shown the model robustness. PSNR is above 51dB for X-ray and MRI medical images and SSIM is 1 without attack. After the Gaussian attack and Geometric attack also the images are not deteriorated. Our proposed method effectively balances image quality and offer a better solution to protect sensitive data.
- 1 December 2025
- Published by The ScienceIn Publishing in Journal of Integrated Science and Technology
- Vol. 14 (1)
Open AccessSecure biometric watermarking in medical images using LH Sub-band Pixel Pair Mapping
Securing a medical image is important and crucial task over digital platform. In this research work, we have proposed a novel method to embed the biometric watermark into medical image. In this process, medical images are more secure for sharing information with medical practitioners over internet. First, we have performed rotation and flipping of medical image. after that Utilization of Discrete Wavelet Transform(DWT) with Haar wavelets decomposes the images into four sub-bands. Watermark is securely embedded in the LH sub band. Now, we have applied Pixel Pair Mapping (PPM) in LH to conceal information bits. Without inverse, the original image can not be retrieved. Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index Measure (SSIM), and Normalized Cross-Correlation (NC) have shown the model robustness. PSNR is above 51dB for X-ray and MRI medical images and SSIM is 1 without attack. After the Gaussian attack and Geometric attack also the images are not deteriorated. Our proposed method effectively balances image quality and offer a better solution to protect sensitive data.
- 22 September 2025
- Published by Springer Nature in International Journal on Document Analysis and Recognition (IJDAR)
- Vol. 29 (2) , 365-380
Devanagari character encoded mix-merge vision transformer for robust document layout analysis
- 22 September 2025
- Published by Springer Nature in International Journal on Document Analysis and Recognition (IJDAR)
- Vol. 29 (2) , 365-380
Devanagari character encoded mix-merge vision transformer for robust document layout analysis
- 14 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 2, 1-5
A Fine-Tuned DenseNet Model for an Efficient Maize Leaf Disease Classification
The detection of maize leaf disease is essential for food security and sustainable agriculture. Early maize leaf disease detection enables timely treatment, limiting crop losses and stopping the replication of diseases. Precise identification facilitates the application of focused interventions, maximizes the use of available resources, and guarantees a more robust yield of maize. In the end, this helps to sustain a steady supply of food and the livelihoods of those who depend on maize farming. This work investigates the use of a DenseNet model for maize leaf disease detection. The model has been supplied with different learning rate values as 0.1, 0.01, 0.001, and 0.0001. The optimizers during the implementation have been also changed to Adam and SGD. The highest accuracy has been achieved by the Adam optimizer at a learning rate of 0.01. The model performs best, demonstrating a 96.53% training accuracy and an even higher 97.43% testing accuracy. The loss of the model has been also analyzed as 0.1433, and 0.0948, during the training and testing phases. The results highlight denseNet’s effectiveness in precisely identifying diseases of the maize leaf, with promising implications for agricultural diagnostics. In the future, the model’s performance can be increased by expanding its identification of a wider variety of maize leaf diseases and combining the model with innovative agricultural technologies to monitor diseases in real-time and react quickly to new challenges associated with leaf diseases.
- 14 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 2, 1-5
A Fine-Tuned DenseNet Model for an Efficient Maize Leaf Disease Classification
The detection of maize leaf disease is essential for food security and sustainable agriculture. Early maize leaf disease detection enables timely treatment, limiting crop losses and stopping the replication of diseases. Precise identification facilitates the application of focused interventions, maximizes the use of available resources, and guarantees a more robust yield of maize. In the end, this helps to sustain a steady supply of food and the livelihoods of those who depend on maize farming. This work investigates the use of a DenseNet model for maize leaf disease detection. The model has been supplied with different learning rate values as 0.1, 0.01, 0.001, and 0.0001. The optimizers during the implementation have been also changed to Adam and SGD. The highest accuracy has been achieved by the Adam optimizer at a learning rate of 0.01. The model performs best, demonstrating a 96.53% training accuracy and an even higher 97.43% testing accuracy. The loss of the model has been also analyzed as 0.1433, and 0.0948, during the training and testing phases. The results highlight denseNet’s effectiveness in precisely identifying diseases of the maize leaf, with promising implications for agricultural diagnostics. In the future, the model’s performance can be increased by expanding its identification of a wider variety of maize leaf diseases and combining the model with innovative agricultural technologies to monitor diseases in real-time and react quickly to new challenges associated with leaf diseases.
- 14 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-7
Feline SentiNet: Deciphering Cat Discomfort with Deep Learning and Random Forest for Pain Grading
In a world where cats' silent suffering is often overlooked, scientists set out on a life-changing journey aided by technology's empathetic hand. The adventure goes beyond human comprehension, venturing into the domain of AI-driven empathy and the search for a key to the secret language of cat agony. They found a way to shed light on the intricacies of feline discomfort using Convolutional Neural Networks (CNNs) and the insight of Random Forest. Each layer in the study reflects a new stanza in the song of recognition, and the research itself evolves like a symphony of pixels. Patterns emerge for the authors when others only see pictures, and they pick up on the agony hidden in the stillness of feline emotions. By bringing together scientific rigor and humaneness, we present a new method for identifying pain in feline companions. Inspiringly, the model distinguished pain intensity among five unique categories with an accuracy of 90%, proving its adeptness in realizing the goals. Researchers saw an instrument that leveled the playing field between man and feline experience when accuracy, recall, and F1 scores all came together. The authors introduce a model that expresses the inexpressible and ease the unspoken as pixels dance, filters learn, and forests decide. In this work, authors celebrate the unspoken connection between people and their feline friends by exploring how technology may serve as a conduit for greater human comprehension, empathy, and the goal of a future in which no one's suffering is ignored.
- 14 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-7
Feline SentiNet: Deciphering Cat Discomfort with Deep Learning and Random Forest for Pain Grading
In a world where cats' silent suffering is often overlooked, scientists set out on a life-changing journey aided by technology's empathetic hand. The adventure goes beyond human comprehension, venturing into the domain of AI-driven empathy and the search for a key to the secret language of cat agony. They found a way to shed light on the intricacies of feline discomfort using Convolutional Neural Networks (CNNs) and the insight of Random Forest. Each layer in the study reflects a new stanza in the song of recognition, and the research itself evolves like a symphony of pixels. Patterns emerge for the authors when others only see pictures, and they pick up on the agony hidden in the stillness of feline emotions. By bringing together scientific rigor and humaneness, we present a new method for identifying pain in feline companions. Inspiringly, the model distinguished pain intensity among five unique categories with an accuracy of 90%, proving its adeptness in realizing the goals. Researchers saw an instrument that leveled the playing field between man and feline experience when accuracy, recall, and F1 scores all came together. The authors introduce a model that expresses the inexpressible and ease the unspoken as pixels dance, filters learn, and forests decide. In this work, authors celebrate the unspoken connection between people and their feline friends by exploring how technology may serve as a conduit for greater human comprehension, empathy, and the goal of a future in which no one's suffering is ignored.
- 2 September 2023
- Published by Springer Nature
- p. 439-449
Deep Learning Based World Coin Currency Detection
- 2 September 2023
- Published by Springer Nature
- p. 439-449
Deep Learning Based World Coin Currency Detection
- 9 September 2022
- Published by Springer Nature
- p. 119-135
An Efficient Regression Test Cases Selection & Optimization Using Mayfly Optimization Algorithm
- 9 September 2022
- Published by Springer Nature
- p. 119-135
An Efficient Regression Test Cases Selection & Optimization Using Mayfly Optimization Algorithm
- 25 July 2022
- Published by Springer Nature in International Journal of Information Technology
- Vol. 15 (2) , 1055-1068
A novel chaotic archimedes optimization algorithm and its application for efficient selection of regression test cases
- 25 July 2022
- Published by Springer Nature in International Journal of Information Technology
- Vol. 15 (2) , 1055-1068
A novel chaotic archimedes optimization algorithm and its application for efficient selection of regression test cases
- 20 September 2021
- Published by Springer Nature
- p. 605-615
Regression Test Case Selection: A Comparative Analysis of Metaheuristic Algorithms
- 20 September 2021
- Published by Springer Nature
- p. 605-615
Regression Test Case Selection: A Comparative Analysis of Metaheuristic Algorithms
- Open Access
- 2 February 2021
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 80 (20) , 30143-30163
Improved optic disc and cup segmentation in Glaucomatic images using deep learning architecture
Glaucoma is an ailment causing permanent vision loss but can be prevented through the early detection. Optic disc to cup ratio is one of the key factors for glaucoma diagnosis. But accurate segmentation of disc and cup is still a challenge. To mitigate this challenge, an effective system for optic disc and cup segmentation using deep learning architecture is presented in this paper. Modified Groundtruth is utilized to train the proposed model. It works as fused segmentation marking by multiple experts that helps in improving the performance of the system. Extensive computer simulations are conducted to test the efficiency of the proposed system. For the implementation three standard benchmark datasets such as DRISHTI-GS, DRIONS-DB and RIM-ONE v3 are used. The performance of the proposed system is validated against the state-of-the-art methods. Results indicate an average overlapping score of 96.62%, 96.15% and 98.42% respectively for optic disc segmentation and an average overlapping score of 94.41% is achieved on DRISHTI-GS which is significant for optic cup segmentation.
- 2 February 2021
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 80 (20) , 30143-30163
Open AccessImproved optic disc and cup segmentation in Glaucomatic images using deep learning architecture
Glaucoma is an ailment causing permanent vision loss but can be prevented through the early detection. Optic disc to cup ratio is one of the key factors for glaucoma diagnosis. But accurate segmentation of disc and cup is still a challenge. To mitigate this challenge, an effective system for optic disc and cup segmentation using deep learning architecture is presented in this paper. Modified Groundtruth is utilized to train the proposed model. It works as fused segmentation marking by multiple experts that helps in improving the performance of the system. Extensive computer simulations are conducted to test the efficiency of the proposed system. For the implementation three standard benchmark datasets such as DRISHTI-GS, DRIONS-DB and RIM-ONE v3 are used. The performance of the proposed system is validated against the state-of-the-art methods. Results indicate an average overlapping score of 96.62%, 96.15% and 98.42% respectively for optic disc segmentation and an average overlapping score of 94.41% is achieved on DRISHTI-GS which is significant for optic cup segmentation.