• KIEAE Journal
• /
• v.16 no.1
• /
• pp.73-79
• /
• 2016

This study aimed to investigate the flood prevention effect expected from the afforestation of a large area metal roof of an industrial complex located in an area prone to floods in the rainwater outflow reduction aspect through computer simulation based on soil, which is a key element of the system. In order to conduct a more realistic simulation, the properties of the surveyed soil were generated through substantive analysis, soil texture analysis, and saxton method. A comparative performance evaluation was conducted by using soil depth and ponding depth, which are key elements of the system, as variables. The study result showed that during the heavy rainfall period, the bottom ash artificial soil had 61% rainwater outflow reduction effect, which was 11% higher than the SWMM standard sand.

• Current Optics and Photonics
• /
• v.7 no.4
• /
• pp.362-377
• /
• 2023

Optical information processing technology is characterized by high speed and parallelism, and the light features short wavelength and large information capacity; At the same time, it has various attributes including amplitude, phase, wavelength and polarization, and is a carrier of multi-dimensional information. Therefore, optical encryption is of great significance in the field of information security transmission, and is widely used in the field of image encryption. For multi-image encryption, this paper proposes a multi-image encryption algorithm based on a modified Gerchberg-Saxton algorithm (MGSA) in the Fresnel-transform domain and computational ghost imaging. First, MGSA is used to realize "one code, one key"; Second, phase function superposition and normalization are used to reduce the amount of ciphertext transmission; Finally, computational ghost imaging is used to improve the security of the whole encryption system. This method can encrypt multiple images simultaneously with high efficiency, simple calculation, safety and reliability, and less data transmission. The encryption effect of the method is evaluated by using correlation coefficient and structural similarity, and the effectiveness and security of the method are verified by simulation experiments.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2022.06a
• /
• pp.854-857
• /
• 2022

본 논문에서는 기존 이미지를 통해 위상 홀로그램을 생성하는 네트워크를 학습 및 최적화하여, 기존에 사용하는 알고리즘 방식인 GS 알고리즘(Gerchberg-Saxton algorithm)을 대체하는 것을 목표로 한다. GS는 반복 최적화 기법으로 한 장의 이미지에서 위상 홀로그램을 생성하는데 많은 시간이 걸리지만, 심층 학습 기반으로 학습된 모델을 통해 위상 홀로그램을 생성할 경우, 반복 최적화 과정 없이 짧은 시간 안에 위상 홀로그램을 생성할 수 있다. GS와 심층 학습 기반으로 각각 생성한 위상 홀로그램을 ASM(Angular Spectrum Method)을 통해 수치적으로 재복원하여 PSNR로 원본 이미지와 비교한 결과, 심층 학습 기반으로 생성한 위상 홀로그램에서 더 좋은 화질의 이미지를 짧은 시간 안에 얻을 수 있었다.