Browsing by Author "Cheng, Chang-Chieh"

Now showing 1 - 10 of 10

Acceleration of Iterative Tomographic Image Reconstruction by Reference-based Back Projection
(2016-01-01)
Cheng, Chang-Chieh
;
Li, Ping-Hui
;
Ching, Yu-Tai
;
資訊工程學系
;
Department of Computer Science
The purpose of this paper is to design and implement an efficient iterative reconstruction algorithm for computational tomography. We accelerate the reconstruction speed of algebraic reconstruction technique (ART), an iterative reconstruction method, by using the result of filtered back-projection (FBP), a wide used algorithm of analytical reconstruction, to be an initial guess and the reference for the first iteration and each back projection stage respectively. Both two improvements can reduce the error between the forward projection of each iteration and the measurements. We use three methods of quantitative analysis, root-mean square error (RMSE), peak signal to noise ratio (PSNR), and structural content (SC), to show that our method can reduce the number of iterations by more than half and the duality of tie. result is better than the original ART.
Correction of center of rotation and projection angle in synchrotron X-ray computed tomography
(2018-06-29)
Cheng, Chang-Chieh
;
Ching, Yu-Tai
;
Ko, Pai-Hung
;
Hwu, Yeukuang
;
資訊工程學系
An error in tomographic reconstruction parameters can result considerable artifacts in the reconstructed image, particularly in micro-computed tomography and nano-computed tomography. This study involved designing an automatic method for efficiently correcting errors resulting from incorrectly determined rotational axes and projection angles. In this method, errors are corrected by minimizing the "total variation" of a reconstructed image, and minimization is accomplished by using the gradient descent method. Compared with two previous methods, the proposed method achieved the best reconstruction results.
Full Model for Sensors Placement and Activities Recognition
(2017-01-01)
Ching, Yu-Tai
;
He, Guan-Wei
;
Cheng, Chang-Chieh
;
Yang, Yu-Jin
;
資訊工程學系
We implemented a wired sensors system that supports activities identification. The system consists of Raspberry Pi, MPU6050 (accelerometers and gyrometers), and TCA9548 (1 to 8 multiplexer). Our experimental results show that when 6 MPU6050 attached to the right arm, right wrist. chest, waist, right thigh, and right ankle, the activities of standing, sitting, lying, walking, running, going upstairs, going downstairs. drinking water, and dumbbells activities could be identified with high accuracy. The system can connect up to 128 sensors, but under a practical sampling rate, the number of sensors should not be greater than 15. The system shall be used for finding the optimal locations for a multi -sensor wearable system (for examples, clothes or shoes).
Handwritten Signature Verification by Using a Six-Axis Motion Sensor and SVM
(2019-01-01)
Cheng, Chang-Chieh
;
Chen, Yi-Chi
;
Ching, Yu-Tai
;
資訊工程學系
;
資訊技術服務中心
Signature verification, if we consider the muscle memory, is a biometric for identification technology. To access muscle memory, we use a motion sensor that consists of accelerometer and gyroscope to implement a signature verification system. The motion sensor records six motion values including three-axis accelerations and angular velocities while name signing. 14 features of signature are extracted from the sequence of accelerations and angular velocities. A support vector machine (SVM) is then applied to verify the signatures. The proposed method was applied to verify the Chinese signatures. The SVM is trained by the training data from each person. The true positive rate of the proposed method can reach to 95.66%. Fake signatures generated by tracing from true signatures can also be recognized by the proposed method.
High-Resolution and Large-Volume Tomography Reconstruction for X-Ray Microscopy
(2016-01-01)
Cheng, Chang-Chieh
;
Hwu, Yeukuang
;
Ching, Yu-Tai
;
資訊工程學系
;
Department of Computer Science
This paper presents a method of X-ray image acquisition for the high-resolution tomography reconstruction that uses a light source of synchrotron radiation to reconstruct a three-dimensional tomographic volume dataset for a nanoscale object. For large objects, because of the limited field-of-view, a projection image of an object should to be taken by several shots from different locations, and using an image stitching method to combine these image blocks together. In this study, the overlap of image blocks should be small because our light source is the synchrotron radiation and the X-ray dosage should be minimized as possible. We use the properties of synchrotron radiation to enable the image stitching and alignment success when the overlaps between adjacent image blocks are small. In this study, the size of overlaps can reach to 15% of the size of each image block. During the reconstruction, the mechanical stability should be considered because it leads the misalignment problem in tomography. We adopt the feature-based alignment method to solve this problem. Finally, we apply the proposed method for a kidney of mouse for verification.
Image Alignment for Tomography Reconstruction from Synchrotron X-Ray Microscopic Images
(2014-01-09)
Cheng, Chang-Chieh
;
Chien, Chia-Chi
;
Chen, Hsiang-Hsin
;
Hwu, Yeukuang
;
Ching, Yu-Tai
A synchrotron X-ray microscope is a powerful imaging apparatus for taking high-resolution and high-contrast X-ray images of nanoscale objects. A sufficient number of X-ray projection images from different angles is required for constructing 3D volume images of an object. Because a synchrotron light source is immobile, a rotational object holder is required for tomography. At a resolution of 10 nm per pixel, the vibration of the holder caused by rotating the object cannot be disregarded if tomographic images are to be reconstructed accurately. This paper presents a computer method to compensate for the vibration of the rotational holder by aligning neighboring X-ray images. This alignment process involves two steps. The first step is to match the "projected feature points" in the sequence of images. The matched projected feature points in the x-theta plane should form a set of sine-shaped loci. The second step is to fit the loci to a set of sine waves to compute the parameters required for alignment. The experimental results show that the proposed method outperforms two previously proposed methods, Xradia and SPIDER. The developed software system can be downloaded from the URL, http://www.cs.nctu.edu.tw/similar to chengchc/SCTA or http://goo.gl/s4AMx.
Poster: Handwritten Signature Verification by Using a Six-Axis Motion Sensor and SVM
(2019-01-01)
Cheng, Chang-Chieh
;
Chen, Yi-Chi
;
Ching, Yu-Tai
;
資訊工程學系
;
資訊技術服務中心
Signature verification, if we consider the muscle memory, is a biometric for identification technology. To access muscle memory, we use a motion sensor that consists of accelerometer and gyroscope to implement a signature verification system. The motion sensor records six motion values including three-axis accelerations and angular velocities while name signing. 14 features of signature are extracted from the sequence of accelerations and angular velocities. A support vector machine (SVM) is then applied to verify the signatures. The proposed method was applied to verify the Chinese signatures. The SVM is trained by the training data from each person. The true positive rate of the proposed method can reach to 95.66%. Fake signatures generated by tracing from true signatures can also be recognized by the proposed method.
Real-time adjustment of transfer function for Fourier volume rendering
(2011-10-01)
Cheng, Chang-Chieh
;
Ching, Yu-Tai
;
資訊工程學系
;
Department of Computer Science
Fourier volume rendering (FVR) is a volume rendering method based on the Fourier slice theorem. With an n x n x n volume data, the FVR algorithm requires O(n(2) log n) time to generate a result. Because it requires time less than O(n(3)) does, FVR is preferred for designing a real-time rendering algorithm with a preprocessing step. We improve upon our previous work. We demonstrate that a B-spline is significantly more useful when designing a transfer function. To de-sign an appropriate transfer function with a spline function, additional control points are required. However, the memory space required for the proposed method increases in linear proportion to the number of control points. We show that the set of control points can be clustered into groups, ensuring the memory required is linearly proportional to the number of groups. The proposed technique supports real-time rendering after adjusting the transfer function for FVR. (C) 2011 SPIE and IS&T. [DOI: 10.1117/1.3653264]
Transfer function design for Fourier volume rendering and implementation using GPU
(2008-01-01)
Cheng, Chang-Chieh
;
Ching, Yu-Tai
;
資訊工程學系
;
Department of Computer Science
Volume rendering is a technique for volume visualization. Given a set of N x N x N volume data, the traditional volume rendering methods generally need O(N-3) rendering time. The FVR (Fourier Volume Rendering), that takes advantage of the Fourier slice theorem, takes O(N-2 log N) rendering time once the Fourier Transform of the volume data is available. Thus the FVR is favor to designing a real-time rendering algorithm with a preprocessing step. But the FVR has a disadvantage that resampling in the frequency domain causes artifacts in the spatial domain. Another problem is that the method for designing a transfer function is not obvious. In this paper, we report that by using the spatial domain zero-padding and tri-linear filtering can reduce the artifacts to an acceptable rendered image quality in spatial domain. To design the transfer function, we present a method that the user can define a transfer function by using a Bezier curve first. Based on the linear combination property of the Fourier transform and Bezier curve equation, the volume rendered result can be obtained by adding the weighted frequency domain signals. That mean, once a transfer function is given, we don't have to recompute the Fourier transform of the volume data after the transfer function applied. This technique makes real-time adjustment of transfer function possible.
Transfer function design for Fourier volume rendering and implementation using GPU - art. no. 691806
(2008)
Cheng, Chang-Chieh
;
Ching, Yu-Tai
;
資訊工程學系
;
Department of Computer Science
Volume rendering is a technique for volume visualization. Given a set of N x N x N volume data, the traditional volume rendering methods generally need O(N-3) rendering time. The FVR (Fourier Volume Rendering), that takes advantage of the Fourier slice theorem, takes O(N-2 log N) rendering time once the Fourier Transform of the volume data is available. Thus the FVR is favor to designing a real-time rendering algorithm with a preprocessing step. But the FVR has a disadvantage that resampling in the frequency domain causes artifacts in the spatial domain. Another problem is that the method for designing a transfer function is not obvious. In this paper, we report that by using the spatial domain zero-padding and tri-linear filtering can reduce the artifacts to an acceptable rendered image quality in spatial domain. To design the transfer function, we present a method that the user can define a transfer function by using a Bezier curve first. Based on the linear combination property of the Fourier transform and Bezier curve equation, the volume rendered result can be obtained by adding the weighted frequency domain signals. That mean, once a transfer function is given, we don't have to recompute the Fourier transform of the volume data after the transfer function applied. This technique makes real-time adjustment of transfer function possible.