Min Beom Lee
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Min Beom Lee has a total of 5 co-authors. Here are the top co-authors by number of shared publications:
- Kang Ryoung Park
Division of Electronics and Electrical Engineering, Dongguk University, 30 Pildong-ro 1-gil, Jung-gu, Seoul 04620, Republic of Korea
3 shared publications - Hyo Sik Yoon
Division of Electronics and Electrical Engineering, Dongguk University, Seoul, South Korea
2 shared publications - Jin Kyu Kang
Division of Electronics and Electrical Engineering, Dongguk University, Seoul, South Korea
2 shared publications - Se Hyun Nam
Division of Electronics and Electrical Engineering, Dongguk University, 30 Pildong-ro, 1-gil, Jung-gu, Seoul 04620, Republic of Korea
1 shared publication - Yu Hwan Kim
Division of Electronics and Electrical Engineering, Dongguk University, 30 Pildong-ro, 1-gil, Jung-gu, Seoul 04620, Korea
1 shared publication
Latest Publications
- Open Access
- 14 January 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 9, 12055-12066
AS-RIG: Adaptive Selection of Reconstructed Input by Generator or Interpolation for Person Re-Identification in Cross-Modality Visible and Thermal Images
Multimodal camera-based person re-identification (ReID) is important in the field of intelligent surveillance. Thermal cameras can solve the problem in that visible-light cameras cannot acquire the valid feature information of a person under poor illumination conditions. However, thermal cameras usually have lower frame resolution than visible-light cameras. To overcome this problem, we propose an adaptive selection of reconstructed input by generator or interpolation (AS-RIG) method, which can adaptively select the generative adversarial network (GAN), or an interpolation method (bi-linear or bi-cubic). AS-RIG automatically selects a resolution-model using the mean-squared error (MSE), feature distance (FD), and structural similarity (SSIM). To verify the performance of our proposed method, two open databases are used: the DBPerson-Recog-DB1 and Sun Yat-set University multiple modality Re-ID (SYSU-MM01). Infrared frames from both databases are resized to be smaller than the original ones for experimentation. Experimental results show that our generator outperforms traditional interpolation methods. In addition, the person ReID experimental results demonstrate that AS-RIG outperforms non-adaptive selection methods and state-of-the-art methods.
- 14 January 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 9, 12055-12066
Open AccessAS-RIG: Adaptive Selection of Reconstructed Input by Generator or Interpolation for Person Re-Identification in Cross-Modality Visible and Thermal Images
Multimodal camera-based person re-identification (ReID) is important in the field of intelligent surveillance. Thermal cameras can solve the problem in that visible-light cameras cannot acquire the valid feature information of a person under poor illumination conditions. However, thermal cameras usually have lower frame resolution than visible-light cameras. To overcome this problem, we propose an adaptive selection of reconstructed input by generator or interpolation (AS-RIG) method, which can adaptively select the generative adversarial network (GAN), or an interpolation method (bi-linear or bi-cubic). AS-RIG automatically selects a resolution-model using the mean-squared error (MSE), feature distance (FD), and structural similarity (SSIM). To verify the performance of our proposed method, two open databases are used: the DBPerson-Recog-DB1 and Sun Yat-set University multiple modality Re-ID (SYSU-MM01). Infrared frames from both databases are resized to be smaller than the original ones for experimentation. Experimental results show that our generator outperforms traditional interpolation methods. In addition, the person ReID experimental results demonstrate that AS-RIG outperforms non-adaptive selection methods and state-of-the-art methods.
- Open Access
- 11 January 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 9, 10120-10135
Enhanced Iris Recognition Method by Generative Adversarial Network-Based Image Reconstruction
Iris recognition is one of the non-contact biometric identification methods that are hygienic and highly accurate. Iris recognition involves using iris images obtained by a near-infrared (NIR) camera or a visible light camera. A clear image of iris can be obtained when an NIR camera is used, but it requires an NIR illuminator in addition to the NIR camera. Iris recognition can be performed with a built-in camera device when a visible light camera is used, which also has the advantage of obtaining a three-channel image containing the color information. Accordingly, studies are being conducted on iris recognition by obtaining iris images from the face images taken by a high-resolution visible light camera in smartphones. However, when iris images have unconstrained conditions or are obtained without the cooperation of the subjects, the quality of iris images are reduced by noises such as optical and motion blur, off-angle view, specular reflection (SR), and other artifacts, thus ultimately deteriorating the recognition performance. Therefore, in this study, a method has been proposed for enhancing the quality of iris images by blurring the iris region and deep-learning-based deblurring. In addition, we propose the method for improving the recognition performance by integrating the recognition score in periocular regions and support vector machine (SVM). The method proposed in this study, which was experimented with noisy iris challenge evaluation-part II training database and MICHE database, exhibited an improved performance compared to the state-of-the-art methods.
- 11 January 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 9, 10120-10135
Open AccessEnhanced Iris Recognition Method by Generative Adversarial Network-Based Image Reconstruction
Iris recognition is one of the non-contact biometric identification methods that are hygienic and highly accurate. Iris recognition involves using iris images obtained by a near-infrared (NIR) camera or a visible light camera. A clear image of iris can be obtained when an NIR camera is used, but it requires an NIR illuminator in addition to the NIR camera. Iris recognition can be performed with a built-in camera device when a visible light camera is used, which also has the advantage of obtaining a three-channel image containing the color information. Accordingly, studies are being conducted on iris recognition by obtaining iris images from the face images taken by a high-resolution visible light camera in smartphones. However, when iris images have unconstrained conditions or are obtained without the cooperation of the subjects, the quality of iris images are reduced by noises such as optical and motion blur, off-angle view, specular reflection (SR), and other artifacts, thus ultimately deteriorating the recognition performance. Therefore, in this study, a method has been proposed for enhancing the quality of iris images by blurring the iris region and deep-learning-based deblurring. In addition, we propose the method for improving the recognition performance by integrating the recognition score in periocular regions and support vector machine (SVM). The method proposed in this study, which was experimented with noisy iris challenge evaluation-part II training database and MICHE database, exhibited an improved performance compared to the state-of-the-art methods.
- Open Access
- 8 November 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 7, 163461-163477
Enhancing the Accuracies of Age Estimation With Heterogeneous Databases Using Modified CycleGAN
- 8 November 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 7, 163461-163477
Open AccessEnhancing the Accuracies of Age Estimation With Heterogeneous Databases Using Modified CycleGAN