Showing papers by "Hyunseok Yang published in 2018"
TL;DR: A fast speckle noise reduction method with only a single reconstructed image based on convolutional neural networks that has multi-sized kernels that can capture the speckled noise component effectively from digital holographic images is proposed.
Abstract: In this Letter, we propose a fast speckle noise reduction method with only a single reconstructed image based on convolutional neural networks. The proposed network has multi-sized kernels that can capture the speckle noise component effectively from digital holographic images. For robust noise reduction performance, the network is trained with a large noisy image dataset that has object-dependent noise and a wide range of noise levels. The experimental results show the fast, robust, and outstanding speckle noise reduction performance of the proposed approach.
55 citations
25 Jun 2018
TL;DR: This work focuses on rapid and efficient estimator to find object distance from hologram in order to reconstruct original image and makes the estimator pre-trained through deep learning.
Abstract: We have focused on rapid and efficient estimator to find object distance from hologram in order to reconstruct original image. Our approach to find it makes the estimator pre-trained through deep learning. Especially in off-axis holography configuration, our method eliminates the unnecessary factors and reduces information loss occurred by resizing image to plug into Convolution Neural Network (CNN). Training is performed on the generated images at several specific distances under various optical conditions and the accuracy of estimation is validated.
4 citations
3 citations
02 Jul 2018
TL;DR: The design and optimal control strategy of an under-actuated, anthropomorphic robotic finger, and a compliance element and series elastic actuators are incorporated in the design to reduce the number of actuators, and for robotic finger control.
Abstract: In this paper, we describe the design and optimal control strategy of an under-actuated, anthropomorphic robotic finger. A compliance element (silicon rubber skin) and series elastic actuators are incorporated in the design to reduce the number of actuators, and for robotic finger control. Manufacturing of the robotic finger is further simplified by the basic joint design and a 3D printing technique. The proposed metacarpophalangeal (MCP) joint structure allows flexion-extension and adduction-abduction motions of the robotic finger to be controlled easily and independently. However, due to the under-actuated system, and joint stiffness caused by the compliance element, the design is limited by inverse kinematics issues, preventing some of the desired joint angles from being achieved. We analyzed this problem and proposed an optimal control strategy to address this issue. Our experimental results showed that good performance of the optimal control strategy of the under-actuated anthropomorphic robotic finger prototype.
1 citations