Publications

Journals

Comprehensive Analysis of Compressible Perceptual Encryption Methods—Compression and Encryption Perspectives
Ijaz Ahmad, Wooyeol Choi and Seokjoo Shin
Sensors, vol. 23, no. 8, p. 4057, Apr. 2023.
[Full Article] Summary

IIB–CPE: Inter and Intra Block Processing-Based Compressible Perceptual Encryption Method for Privacy-Preserving Deep Learning
Ijaz Ahmad and Seokjoo Shin
Sensors, vol. 22, no. 20, p. 8074, Oct. 2022.
[Full Article] Summary

A Perceptual Encryption-Based Image Communication System for Deep Learning-Based Tuberculosis Diagnosis Using Healthcare Cloud Services
Ijaz Ahmad and Seokjoo Shin
Electronics, vol. 11, no. 16, p. 2514, Aug. 2022.
[Full Article] Summary

A Novel Hybrid Image Encryption–Compression Scheme by Combining Chaos Theory and Number Theory
Ijaz Ahmad and Seokjoo Shin
Signal Processing: Image Communication, vol. 98, p. 116418, Oct. 2021.
[Full Article] Summary

Description: Data transmission has dual requirements of compression and encryption.… The order in which the two operations are carried out affects the overall efficiency of multimedia application. The encrypted data has less or no compressibility while, it is challenging to ensure reasonable security without downgrading the compression performance. Therefore, incorporating one requirement into another is an interesting approach. We propose a novel hybrid image encryption and compression scheme that allows compression in the encryption domain. The encryption is based on Chaos theory and is carried out in two steps, i.e., permutation and substitution. The lossless compression is performed on the shuffled image and then the compressed bitstream is grouped into 8-bit elements for substitution stage. In addition, we also porpose a novel data-to-symbol mapping method based on Chinese Remainder Theorem for the improved compression savings. + Read more

---

International Conferences

Perceptual Image Encryption: A Communication Perspective
Ijaz Ahmad and Seokjoo Shin
IEEE International Conference on Information Networking (ICOIN): xxxx–xxxx, Jan. 2024.
[Full Article] Summary

Description: This study integrated perceptual encryption algorithms into the source coding block of an orthogonal frequency-division multiplexing …(OFDM) system and analyzed their performance in terms of the recovered image quality. + Read more

Deep Learning-based Image Quality Assessment Metric for Quantifying Perceptual Distortions in Transmitted Images
Ijaz Ahmad and Seokjoo Shin
IEEE International Symposium on Communications and Information Technologies (ISCIT): 208-213, Oct. 2023.
[Full Article] Summary

Description: We proposed a convolutional autoencoder-based image quality assessment metric …that compares images in low dimensional feature space and can be used to monitor image degradation occurred during data transmission. + Read more

Noise-cuts-Noise Approach for Mitigating the JPEG Distortions in Deep Learning
Ijaz Ahmad and Seokjoo Shin
IEEE International Conference on Artificial Intelligence in Information and Communication (ICAIIC): 221–226, Feb. 2023.
[Full Article] Summary

Description: To improve the classification performance of the trained deep learning models… (inference phase) on the JPEG compressed images, we propose a noise-based data augmentation technique called noise-cuts-noise approach. + Read more

Robustness of Deep Learning enabled IoT Applications Utilizing Higher Order QAM in OFDM Image Communication System
Nazmul Islam, Ijaz Ahmad and Seokjoo Shin
IEEE International Conference on Artificial Intelligence in Information and Communication (ICAIIC): 630–635, Feb. 2023.
[Full Article] Summary

Description: Image data transmission over a wireless communication channel is susceptible to distortions…. We analyze the robustness of the trained deep learning models on such images where the noise is introudced by an image communication system that utilizes higher order M-QAM schemes. + Read more

Perceptual Encryption-based Privacy-Preserving Deep Learning for Medical Image Analysis
Ijaz Ahmad and Seokjoo Shin
IEEE International Conference on Information Networking (ICOIN): 224–229, Jan. 2023.[BEST PAPER AWARD]
[Full Article] Summary

Description: When the image data are outsourced to avail third-party owned resources… such as cloud computation services, there is a privacy concern that the data should be kept protected not only during transmission but during computations as well. To meet these requirements, we proposed a privacy-preserving Deep Learng (PPDL) scheme based on perceptual encryption, that enable computations without the need of decryption. + Read More

Performance Analysis of Cloud-based Deep Learning Models on Images Recovered without Channel Correction in OFDM System
Ijaz Ahmad, Nazmul Islam and Seokjoo Shin
IEEE Asia Pacific Conference on Communication (APCC): 255-259, Oct. 2022.
[Full Article] Summary

Description: Channel correction plays an important role in performance of wireless communication systems.…In this study, we considered the otherwise to bypass channel estimation of an Orthogonal Frequency Division Multiplexing (OFDM) based image communication system designed to enable cloud-based deep learning (DL) computation . + Read More

Perceptual Encryption-based Privacy-Preserving Deep Learning in Internet of Things Applications
Ijaz Ahmad and Seokjoo Shin
IEEE International Conference on Information and Communication Technology Convergence (ICTC): 1817–1822, Oct. 2022.
[Full Article] Summary

Description: For an efficient edge-cloud collaboaration to enable AI-based IoT applications …, the trained models should be periodically trained on the most recent data. This necessitates teh data share between cloud and edge servers. However, the data acquired by IoT end devices usually consists of sensitive information and sharing them with cloud services provider results in users’ privacy concern. To this end, we propose perceptual encryption-based method to enable DL computations in encryption domain in order to protect users’ privacy. + Read More

Quantitative Assessment of the Impact of Lossy JPEG Compression on Deep Learning Models
Ijaz Ahmad and Seokjoo Shin
KINGPC International Conference on Next Generation Computing (ICNGC): 249-252, Oct. 2022.
[Full Article] Summary

Description: The design of lossy algorithms is based on a principle to discard information that are not perceivable by human visual system (HVS). … With the popularity of deep learning models (DL) in computer vision (CV), it is necessary to characterize the loss in image quality with respect to computer vision systems as well. We present a quantitative assessment of different types of data loss that occurs due to chroma subsampling, quantization, and rounding functions of the JPEG algorithm. + Read More

Privacy-Preserving Surveillance for Smart Cities
Ijaz Ahmad, Eunkyung Kim, Suk-Seung Hwang and Seokjoo Shin
IEEE International Confernce on Ubiquitous and Future Networks (ICUFN): 301-306, Jul. 2022.
[Full Article] Summary

Description: We present an application of perceptual encryption for privacy-preserving face recognition task. … The main advantage of the proposed method is that it retains color information and the encryption has no effect on the ML algorithm accuracy. + Read More

Determining Jigsaw Puzzle State from an Image based on Deep Learning
Ijaz Ahmad, Suk-Seung Hwang and Seokjoo Shin
IEEE International Conference on Artificial Intelligence in Information and Communication (ICAIIC): 030–032, Feb. 2022.
[Full Article] Summary

Description: We proposed a deep learning-based method to determine state of the jigsaw puzzle in an input image.… The task is represented as a classification problem where each state of the puzzle is considered as a class. As an application, we have shown that the trained model can be used as a fitness function of genetic algorithm-based jigsaw puzzle solver. + Read More

Encryption-then-Compression System for Cloud-based Medical Image Services
Ijaz Ahmad and Seokjoo Shin
IEEE International Conference on Information Networking (ICOIN): 30–33, Jan. 2022. [BEST PAPER AWARD]
[Full Article] Summary

Description: We extend the applications of perceptual encryption-based encryption-then-compression (EtC) schemes to medical image processing domain … to satisfy the dual requirement of a communication system (that is compression and encryption). Given the importance of medical images data, we propose a method to improve the security of the conventional EtC methods. We have implemented a deep learning model for tuberculosis screeining in Montgomery datasets. + Read More

Fine-Tuning Pre-Trained Deep Learning Models for Multiclass Grayscale Images Classification
Ijaz Ahmad and Seokjoo Shin
KINGPC International Conference on Next Generation Computing (ICNGC): 244-246, Nov. 2021.
[Full Article] Summary

Description: We extend our ICAIIC 2021’s approach to run pre-trained models on grayscale images … while benefiting from transfer learning for multiclass classification problem. + Read More

An Approach to Run Pre-Trained Deep Learning Models on Grayscale Images
Ijaz Ahmad and Seokjoo Shin
IEEE International Conference on Artificial Intelligence in Information and Communication (ICAIIC): 177–180, Apr. 2021.
[Full Article] Summary

Description: Grayscale images have smaller sizes than color images thereby, can enable real time computer vision applications … in scenarios where there is constraint on device memory and available bandwidth. We propose an approach to run pre-trained models on grayscale images for binary classification task. + Read More

Region-based Selective Compression and Selective Encryption of Medical Images
Ijaz Ahmad and Seokjoo Shin
ACM International Conference on Smart Media and Applications (SMA): 34–38, Sep. 2020. [BEST PAPER AWARD]
[Full Article] Summary

Description: We propose a method for selective compression and selectinve encryption of medical images. … It is based on lossless compression and encryption of the region-of-interest (ROI) in medical images. The non-ROI part of the image is compressed in lossy mode and is stored or transmitted as plain data, in order to further reduce the image size and to avoid the computational cost of encrypting huge volumes of medical images. + Read More

Analysis of Chinese Remainder Theorem for Data Compression
Ijaz Ahmad, Bumshik Lee and Seokjoo Shin
IEEE International Conference on Information Networking (ICOIN): 634–636, Jan. 2020.
[Full Article] Summary

Description: Chinese remainder theorem (CRT) forms the foundation of recent simultaneous encryption and compression schemes for image data. … We analyze the number of moduli (i.e. block size) and their values effect on the compression performance of CRT methods. + Read More

Just-Noticeable-Difference Based Edge Map Quality Measure
Ijaz Ahmad and Seokjoo Shin
KINGPC International Conference on Next Generation Computing (ICNGC): Dec 20-22, 2018.[BEST POSTER AWARD]
[Full Article] Summary

Description: We propose edge map quality measure based on Just-Noticeable-Difference (JND) feature of human visual system …, to compensate the shortcomings of distance-based edge measures. For this purpose, we design constant stimulus experiment to measure the JND value for two spatial alternative. . + Read More

Color-to-grayscale algorithms effect on edge detection—A comparative study
Ijaz Ahmad, Inkyu Moon and Seokjoo Shin
IEEE International Conference on Electronics, Information, and Communication (ICEIC): 1-4, Jan. 2018.
[Full Article] Summary

Description: We evaluate the impact of various color-to-grayscale algorithms in edge detection. … We implement ten different color-to-grayscale conversion algorithms and the resultant grayscale images were further tested with eight different edge detection algorithms. + Read More

---

National Conferences

Learnable Pixel-Based Encryption for Privacy-Preserving Image Classification
Korean Title: 개인정보 보호 이미지 분류를 위한 학습 가능한 픽셀 기반 암호화
Ijaz Ahmad and Seokjoo Shin
KICS Winter Conference (동계종합학술발표회): 252-253 Jan. 2024.
[Full Article]

Chroma Subsampling for Sub-block-based Perceptual Encryption Algorithms
Korean Title: 서브 블록기반 지각 암호화 알고리즘을 위한 크로마 서브샘플링
Ijaz Ahmad and Seokjoo Shin
KICS Fall Conference (추계종합학술발표회): 852-853, Nov. 2023.
[Full Article]

Deep Joint Source-Channel Coding for Medical Image Transmission
Korean Title: 의료영상 전송을 위한 딥 조인트 소스-채널 코딩
Ijaz Ahmad and Seokjoo Shin
KICS Korea Artificial Intelligence Conference (제4회 한국 인공지능 학술대회): 449-450, Sep. 2023.
[Full Article]

Convolutional Autoencoder for Image Quality Assessment
Korean Title: 합성곱 오토인코더 기반의 이미지 품질 평가
Ijaz Ahmad and Seokjoo Shin
KICS Summer Conference (하계종합학술발표회): 1304-1305, Jun. 2023.
[Full Article]

Leveraging Transfer Learning in EfficientNetV2-based Tuberculosis Detection
Korean Title: EfficientNetV2 기반 결핵 검출을 위한 전이 학습 활용
Ijaz Ahmad and Seokjoo Shin
KICS Fall Conference (추계종합학술발표회): 458-459, Nov. 2022.
[Full Article]

A Comparison of EfficientNets for Tuberculosis Detection in Chest Radiographs
Korean Title: 흉부 엑스선 결핵 검출을 위한 EfficientNets 비교
Ijaz Ahmad and Seokjoo Shin
KICS Korea Artificial Intelligence Conference (제3회 한국 인공지능 학술대회): 126-127, Sep. 2022.
[Full Article]

Performance Analysis of Cloud based Deep Learning Models in OFDM based Image Communication system
Korean Title: OFDM 기반 영상통신 시스템의 클라우드 기반 딥러닝 모델의 성능분석에 관한 연구
Ijaz Ahmad, Nazmul Islam, Eunkyung Kim and Seokjoo Shin
KICS Summer Conference (하계종합학술발표회): 0500-0501, Jun. 2022.
[Full Article]

Block-based Perceptual Encryption Algorithm with Improved Color Components Scrambling
Korean Title: 향상된 색상 구성요소 스크램블링을 사용한 블록 기반 지각 암호화 알고리즘
Ijaz Ahmad and Seokjoo Shin
KINGPC Spring Conference (춘계학술대회): 155-158, May. 2022.
[Full Article]

A Pixel-based Encryption Method for Privacy-Preserving Deep Learning Models
Ijaz Ahmad and Seokjoo Shin
KICS Winter Conference (동계종합학술발표회): 549-551, Feb. 2022.
[Full Article]

Optimal Batch Size for Fine-Tuning Pre-Trained Deep Learning Models
Korean Title: 사전 훈련된 딥 러닝 모델의 파인튜닝을 위한 최적의 배치 크기
Ijaz Ahmad and Seokjoo Shin
KICS Korea Artificial Intelligence Conference (제2회 한국 인공지능 학술대회): 203-205, Sep. 2021.
[Full Article]

Effect of Inter and Intra Block-level Shuffling on the JPEG Compression Performance
Korean Title: 인터 및 인트라 블록 수준 셔플링이 JPEG 압축 성능에 미치는 영향
Ijaz Ahmad and Seokjoo Shin
한국정보과학회 Summer Worshop on Computer Communication (하계 컴퓨터통신 워크샵): 45-48, Aug. 2021.
[Full Article]

Optimal Resolution Selection to Run Pre-Trained Deep Learning Models on Tiny Images
Ijaz Ahmad and Seokjoo Shin
KINGPC Spring Conference (춘계학술대회): 293-295, May. 2021. [BEST POSTER AWARD]
[Full Article]

Data Rate of End-to-End Learning of Communication Systems A Survey
Ijaz Ahmad and Seokjoo Shin
KINGPC Summer Conference(하계학술대회): 1-4, Aug. 2020.
[Full Article]

Performance analysis of Chinese Remainder Theorem for Data Compression
Ijaz Ahmad and Seokjoo Shin
한국정보과학회 Korean Computing Conference(한국컴퓨터종합학술대회): 1101-1103, Jul. 2020.
[Full Article]

Channel Model for End-to-End Learning of Communications Systems: A Survey
Ijaz Ahmad and Seokjoo Shin
KISM Spring Conference (한국스마트미디어학회 춘계학술대회): 209-212, May. 2020.
[Full Article]

Analysis of Chinese Remainder Theorem Moduli for Image Compression
Ijaz Ahmad and Seokjoo Shin
KICS Fall Conference(추계종합학술발표회): 108-109, Nov. 2019.
[Full Article]

Algorithmic Design of Korean Dancheong Patterns
Korean Title: 한국 단청 패턴 알고리즘 설계
Ijaz Ahmad, Najma Sahar and Seokjoo Shin
KISM Workshop on Covergent and Smart Media Systems (융합∙스마트미디어시스템 워크샵): 35-38, Jan. 2019.
[Full Article]

---