• Journal of the Korean Institute of Telematics and Electronics S
• /
• v.35S no.7
• /
• pp.95-102
• /
• 1998

This paper proposes a synchronization method of synthetic facial iamge sequences and synthetic speech. The LP-PSOLA synthesizes the speech for each demi-syllable. We provide the 3,040 demi-syllables for unlimited synthesis of the Korean speech. For synthesis of the Facial image sequences, the paper defines the total 11 fundermental patterns for the lip shapes of the Korean consonants and vowels. The fundermental lip shapes allow us to pronounce all Korean sentences. Image synthesis method assigns the fundermental lip shapes to the key frames according to the initial, the middle and the final sound of each syllable in korean input text. The method interpolates the naturally changing lip shapes in inbetween frames. The number of the inbetween frames is estimated from the duration time of each syllable of the synthetic speech. The estimation accomplishes synchronization of the facial image sequences and speech. In speech synthesis, disk memory is required to store 3,040 demi-syllable. In synthesis of the facial image sequences, however, the disk memory is required to store only one image, because all frames are synthesized from the neutral face. Above method realizes synchronization of system which can real the Korean sentences with the synthetic speech and the synthetic facial iage sequences.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.3 no.1
• /
• pp.8-18
• /
• 2008

In this paper, we propose an efficient algorithm using wavelet transform for still image database retrieval. Especially, it uses only the lowest frequency sub-band in multi-level wavelet transform so that a retrieval system uses a smaller quantity of memory and takes a faster processing time. We extract different textured features, statistical information such as mean, variance and histogram, from low frequency sub-band. Then we measure the distances between the query image and the images in a database in terms of these features. To obtain good retrieval performance, we use the first feature (mean and variance of wavelet coefficients) to filter out most of the unlikely images. The rest of the images are considered to be candidate images. Then we apply the second feature (histogram of wavelet coefficient) to rank all the candidate images. To evaluate the algorithm, we create various distorted image databases using MIT VisTex texture images and PICS natural images. Through simulations, we demonstrate that our method can achieve performance satisfactorily in terms of the retrieval accuracy as well as the both memory requirement and computational complexity. Therefore it is expected to provide good retrieval solution for JPEG-2000 using wavelet transform.

• Journal of Biomedical Engineering Research
• /
• v.21 no.5
• /
• pp.527-535
• /
• 2000

For a reliable patient set-up verification, better portal films are needed to track relevant features. Simulator films are compared with portal films as a reference image in radiotherapy planning. This shows some possibilities of the use of image information of simulator images for enhancement and restorations of portal images which are very poor in quality compared with the simulator images. This paper present an approach that combine an associative memory, a kind of artificial neural networks with fuzzy image enhancement technique using genetic algorithm which determines the fuzzy region of membership function by the use of maximum entropy principles. A higher portal image quality than conventional technique is achieved.

• Proceedings of the Korean Institute of Communication Sciences Conference
• /
• 1987.04a
• /
• pp.31-34
• /
• 1987

A new efficientcoding system which can transmit video conferenceof viedeophone signals at a 64kbps is proposed. In addition to the interframe and CRC (Conditional Repleni shment Coding) system, BTVQ (Binary Tree searched Vector Quantizer)and RLC (Run Length Coding) methods are incorporated. Couble buffer memory is used for simple comtrol of channel symbol transmission and memory underflow And also buffer memory onerfolw is easily controlled by the thresholds of a MAD (Moving Area Betector)

• Proceedings of the KIEE Conference
• /
• 1988.07a
• /
• pp.247-249
• /
• 1988

The most space-consuming element of digital image processing system is the video memory. Though this problem is solved by DRAMs, timing constraints posed by video data rates. The cycle time of DRAMs can be diminished by serial transferring and reading or writing pixel datas at the same time. This paper resents the design of 1024${\times}$512 video memory using this technique.

• Journal of KIISE:Computer Systems and Theory
• /
• v.31 no.3_4
• /
• pp.187-194
• /
• 2004

Shear-Warp volume rendering has many advantages such as good image Quality and fast rendering speed. However in the interactive classification environment it has low efficiency of memory access since preprocessed classification is unavailable. In this paper we present an algorithm using the spacial locality of memory access in the interactive classification environment. We propose an extension model appending a rotation matrix to the factorization of viewing transformation, it thus performs a scanline-based rendering in the object and image space. We also show causes and solutions of three problems of the proposed algorithm such as inaccurate front-to-back composition, existence of hole, increasing computational cost. This model is efficient due to the spacial locality of memory access.

• Smart Structures and Systems
• /
• v.3 no.1
• /
• pp.89-114
• /
• 2007

Two aspects of the design of a small-scale smart wing are addressed in this work, related to the ability of the wing to modify its cross section assuming the shape of two different airfoils and to the possibility of deflecting the profiles near the trailing edge in order to obtain hingeless control surfaces. The actuation is provided by one-way shape memory alloy wires eventually coupled to springs, Shape Memory Alloys (SMAs) being among the most promising materials for this kind of applications. The points to be actuated along the profiles and the displacements to be imposed are selecetd so that they satisfactorily approximate the change from an airfoil to the other and to result in an adequate deflection of the control surface; the actuators and their performances are designed so that an adequate wing stiffness is guaranteed, in order to prevent excessive deformations and undesired airfoil shape variations due to aerodynamic loads. The effect of the pressure distributions, calculated by way of the XFOIL software, and of the actuators loads, is estimated by FE analyses of the loaded wing. Two prototypes are then realised incorporating the variable airfoil and the hingeless aileron features respectively, and the verification of their shapes in both the actuated and non-actuated states, supported by image analysis techniques, confirms that interesting results are achievable with the proposed lay out and design considerations.

• Smart Media Journal
• /
• v.10 no.1
• /
• pp.63-69
• /
• 2021

When a person sees a sentence and understands the sentence, the person understands the sentence by reminiscent of the main word in the sentence as an image. Text-to-image is what allows computers to do this associative process. The previous deep learning-based text-to-image model extracts text features using Convolutional Neural Network (CNN)-Long Short Term Memory (LSTM) and bi-directional LSTM, and generates an image by inputting it to the GAN. The previous text-to-image model uses basic embedding in text feature extraction, and it takes a long time to train because images are generated using several modules. Therefore, in this research, we propose a method of extracting features by using the attention mechanism, which has improved performance in the natural language processing field, for sentence embedding, and generating an image by inputting the extracted features into the GAN. As a result of the experiment, the inception score was higher than that of the model used in the previous study, and when judged with the naked eye, an image that expresses the features well in the input sentence was created. In addition, even when a long sentence is input, an image that expresses the sentence well was created.

• Proceedings of the IEEK Conference
• /
• 1998.10a
• /
• pp.727-730
• /
• 1998

In this paper, we propose an implementation of an efficient parallel Radon transform on TMS320C80-based system. For an N$\times$N SAR image, we can obtain O(NM/p) of the conventional parallel Radon transform, by representing the projection patterns in Radon space variables instead of the image space variables, and pipelining the algorithm, where p is the number of processors and M is the number of projection angles. Also, we can reduce the time for the dynamic load distribution among the nodes and the communication overheads of accessing the global memories, by pipelining the memory and processing operations by using tripple buffer structure. Experimental results show an efficient parallel Radon transform of speedup Sp=3.9 and efficiency E=97.5% for 256$\times$256 image, when implemented on TMS320C80 composed of four parallel slave processors with three memory blocks.

• Journal of Biomedical Engineering Research
• /
• v.12 no.2
• /
• pp.107-112
• /
• 1991

In this paper, we have developed page level input system of the character reading aid for the blind. Input toys)ems arse consisted with 512 pixels line image sensor, optical lento, digital interface for the computer and its control software. Input buffer size of the computer memory that for the single scanning of printed matters Image is 64kB. Image patterns of the reading characters which stored in system memory are converted to tactile character patterns that would be output to the bimorph tactile sensor by software control.