Browsing by Author "Syu, Jia-Hao"

Now showing 1 - 7 of 7

A PATCH-BASED CONTRACTION PROCESS FOR THE IMPROVEMENT OF IMAGE MATTING
(2017-01-01)
Syu, Jia-Hao
;
Wang, Sheng-Jyh
;
Wang, Li-Chun
;
電子工程學系及電子研究所
;
電信工程研究所
In this paper, we propose a patch-based contraction process for the improvement of image matting. Given an input image and a trimap, the proposed contraction process quickly refines the trimap by reclassifying the alpha values of the undetermined pixels into "foreground", "background", or "undetermined", based on the feature similarity of the undetermined pixels with respect to the foreground/background regions. The refinement of trimap causes the shrinkage of undetermined regions and may reveal more foreground/background information across isolated regions. Based on the refined trimap, the image matting algorithm can better estimate the alpha values of the image. Experimental results show that the proposed patch based process provides an effective and efficient way to obtain a refined trimap and the use of the trimap refinement improves the performance of image matting.
Conversion of absorption to fluorescence probe in solid-state sensor for nitric oxide and nitrite
(2013-04-01)
Hong, Wun-Yan
;
Yang, Tsung-Wei
;
Wang, Chun-Ming
;
Syu, Jia-Hao
;
Lin, Yu-Chao
A hydrogel bi-layer structure is used to convert an absorption type to a more convenient fluorescent type probe film for aqueous nitric oxide (NO) and nitrite (NO2-). In the bottom layer the commercial probe molecule 1,2-diaminoanthraquinone (DAQ) is dispersed in poly 2-hydroxyethyl methacrylate (poly HEMA), in the top layer the red inorganic phosphor is dispersed in poly HEMA. Both optical excitation and detection is from the bottom. This bi-layer is integrated with organic light-emitting diode (OLED) and organic photo-detector to form a real-time solid-state sensor. The sensor is responsive to NO bubbling, and remains stable in acid condition. Crown Copyright (C) 2012 Published by Elsevier B.V. All rights reserved.
Electrospun Fibers as a Solid-State Real-Time Zinc Ion Sensor with High Sensitivity and Cell Medium Compatibility
(2013-03-25)
Syu, Jia-Hao
;
Cheng, Yi-Kai
;
Hong, Wun-Yan
;
Wang, Hsing-Ping
;
Lin, Yu-Chao
meso-2,6-Dichlorophenyltripyrrinone (TPN-Cl2), a probe molecule for zinc II ions, is dispersed in a polymer host. The red fluorescence peak at 620 nm appears when the molecule forms a complex with zinc at its center. TPN-Cl2 has a high selectivity for zinc II and tolerates many common metal ions present in the human body. The probe molecules are blended with a hydrogel polymer, poly(2-hydroxyethyl methacrylate) (poly HEMA), with 30 wt% dimethylformamide (DMF). The fiber structure with 1 m diameter is made by electrospinning in DMF solution of the probe and poly HEMA mixture. The fibrous film detects zinc ions with concentrations as low as 106 M in real-time both in water and in the commonly used cell culture liquid media Dulbecco's modified Eagle medium (DMEM) and fetal bovine serum (FBS), which contain many metal ions and proteins. The time-resolution is 5 min for 106 M and 1 min for 105 M. This sensitivity and response speed satisfy the requirements for non-invasive biomedical studies.
Hierarchical Image Segmentation Based on Iterative Contraction and Merging
(2017-05-01)
Syu, Jia-Hao
;
Wang, Sheng-Jyh
;
Wang, Li-Chun
;
電機學院
;
電信工程研究所
In this paper, we propose a new framework for hierarchical image segmentation based on iterative contraction and merging. In the proposed framework, we treat the hierarchical image segmentation problem as a sequel of optimization problems, with each optimization process being realized by a contraction-and-merging process to identify and merge the most similar data pairs at the current resolution. At the beginning, we perform pixel-based contraction and merging to quickly combine image pixels into initial region-elements with visually indistinguishable intra-region color difference. After that, we iteratively perform region-based contraction and merging to group adjacent regions into larger ones to progressively form a segmentation dendrogram for hierarchical segmentation. Comparing with the state-of-the-art techniques, the proposed algorithm can not only produce high-quality segmentation results in a more efficient way, but also keep a lot of boundary details in the segmentation results.
Image Enhancement based on Re gion-based Contraction Process
(2018-01-01)
Syu, Jia-Hao
;
Wang, Sheng-Jyh
;
Wang, Li-Chun
;
電子工程學系及電子研究所
;
電信工程研究所
In this paper, we propose a region-based contraction process for image enhancement. At the beginning, the input image is converted into HSV color space to preserve the hue and saturation information. The brightness V is separated into 256 levels of regions. After that, we add in two pseudo regions with the brightness values 0 and 255 and our image enhancement algorithm pulls the brightness of image regions forward 0 or 255 based on a region-based contraction process. Experimental results show that the proposed process provides an effective and efficient way to implement image enhancement.
SEMANTIC SEGMENTATION BASED ON ITERATIVE CONTRACTION AND MERGING
(2017-01-01)
Yang, Tzu-Hao
;
Syu, Jia-Hao
;
Wang, Sheng-Jyh
;
交大名義發表
;
National Chiao Tung University
The state-of-the-art models for semantic image segmentation usually contain a convolutional neural network (CNN) and a conditional random field (CRF). As a predictor, existing CNN techniques can generate a dense prediction result but may generate obvious boundary errors at the same time. As a refinement model, CRF improves the CNN outcomes by forcing the consistency of local labels. However, the use of CRF may cause fragmentation effect around object boundaries. In this paper, we propose the use of a so-called iterative contraction and merging (ICM) process to facilitate the semantic segmentation process. Guided by the high-level information from CNN, the ICM process is used as a tool to grow image segments in a bottom-up way and to produce more accurate outcomes in an iterative way. The ICM process can faithfully preserve the boundary information and maintain the consistency of local labels. Our experimental results demonstrate that the performance of the proposed approach is comparable to the state-of-the-art models but with more accurate boundaries.
Semantic Segmentation of Indoor-Scene RGB-D Images Based on Iterative Contraction and Merging
(2018-01-01)
Syu, Jia-Hao
;
Cho, Shih-Hsuan
;
Wang, Sheng-Jyh
;
Wang, Li-Chun
;
電子工程學系及電子研究所
In this paper, we propose an iterative contraction and merging framework (ICM) for semantic segmentation in indoor scenes. Given an input image and a raw depth image, we first derive the dense prediction map from a convolutional neural network (CNN) and a normal vector map from the depth image. By combining the RGB-D image with these two maps, we can guide the ICM process to produce a more accurate hierarchical segmentation tree in a bottom-up manner. After that, based on the hierarchical segmentation tree, we design a decision process which uses the dense prediction map as a reference to make the final decision of semantic segmentation. Experimental results show that the proposed method can generate much more accurate object boundaries if compared to the state-of-the-art methods.