• Proceedings of the Korea Multimedia Society Conference
• /
• 2002.05c
• /
• pp.257-262
• /
• 2002

영상을 적은 비트로 표현할 때 먼저 양자화를 이용하여 칼라맵을 생성한다. 그리고 적은 비트의 칼라맵으로도 인간의 시각에 적합하게 표현하기 위해 디더링을 결합한다. 본 논문에서는 디더링 기법중 오차확산법이 주변화소로 양자화 에러를 확산한다는 것을 고려하여 칼라맵을 생성하는 새로운 방법을 제안한다. 제안방법은 LBG 알고리즘의 개선하여 클러스터의 양자화 벡터를 구하는 각각의 반복단계에서 현재 양자화 벡터와 새로운 중심값(centroid)을 연결하는 직선 상에서 새로운 양자화벡터를 구하는 기존의 알고리즘에 에러를 고려하여 새로운 양자화 벡터를 얻을 수 있도록 하였다. 제안방법을 적용하였을 때 기존의 LBG 알고리즘에 비해 양자화 영상과 디더영상의 화질이 개선되었다. 또한 각 칼라별 MSE 와 영상전체 MSE 에 대해서도 제안방법은 기존의 LBG 알고리즘에 대해 개선되었다.

• Journal of the Korean Institute of Telematics and Electronics
• /
• v.23 no.2
• /
• pp.249-256
• /
• 1986

In this paper, we propose the MDM algorithm by which one can desing an FIR filter that is maximally flat and requires no multiplication. We use the modified MAXFLAT subroutive of Kaiser to achieve the maximally-flat characteristics. The filter coefficients are encoded in MDM-code and the optimal stepsize is determined the steepest- descent method. Simulation results shows that the FIR filter designed is almost maximally-flat in passband, but has about -30dB ripples in stopband due to MDM quantization error.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.8 no.4
• /
• pp.561-567
• /
• 2013

Some of the feature vectors for speech recognition are rare and unusual. These patterns lead to overfitting for the parameters of the speech recognition system and, as a result, cause structural risks in the system that hinder the good performance in recognition. In this paper, as a method of removing these unusual patterns, we try to exclude vectors whose norms are larger than a specified cutoff value and then train the speech recognition system. The objective of this study is to exclude as many unusual feature vectors under the condition of no significant degradation in the speech recognition error rate. For this purpose, we introduce a cutoff parameter and investigate the resultant effect on the speaker-independent speech recognition of isolated words by using FVQ(Fuzzy Vector Quantization)/HMM(Hidden Markov Model). Experimental results showed that roughly 3%~6% of the feature vectors might be considered as unusual, and therefore be excluded without deteriorating the speech recognition accuracy.

• The KIPS Transactions:PartB
• /
• v.11B no.6
• /
• pp.625-632
• /
• 2004

Since the H.264 encoding adopts both forward prediction and hi-direction prediction modes and exploits Variable Length Coding(VLC), the amount of data generated from video encoder varies as Flaying time goes by. The fixed bit rate encoding system which has limited transmission channel capacity uses a buffer to control output bitstream It's necessary to control the bitstream to maintain within manageable range so as to protect buffer from overflow or underflow. With existing bit amount control algorithms, the $\lambda_{MODE}$ which is relationship between distortion value and quantization parameter often excesses normal value to end up with video error. This paper proposes an algorithm to protect buffer from overflow or underflow by introducing a new quantization parameter against distortion value of H.264 video data. The test results of 6 exemplary data show that the proposed algorithm has the same PSNR as and up to 8% reduced bit rate against existing algorithms.

• Journal of Broadcast Engineering
• /
• v.11 no.2 s.31
• /
• pp.191-199
• /
• 2006

This paper presents an efficient pre echo control scheme for MPEG Audio coders based on the psychoacoustic model II (PAM-II). Pre echo control is the final step for the calculation of masking threshold in the PAM II. It is to minimize the spread of quantization error over the processing frame. In the conventional encoders, pre echo is reduced by restricting the estimated masking threshold not to exceed the one obtained in the previous frame. The conventional method performs pre echo control not only for short blocks but also for long blocks, which lowers the masking threshold in long blocks and, in turn, increases the quantization noise level of corresponding blocks. This paper proposes an efficient pre echo control process. The test result shows a mean enhancement of more than 0.4 especially for complex signals on the ITU R 5 point audio impairment scale.

• The Transactions of the Korea Information Processing Society
• /
• v.3 no.5
• /
• pp.1307-1316
• /
• 1996

In this paper, we propose a vector quantization method which uses a dynamic address mapping based on exploring the high interblock correlation. In the proposed method, we reduce bit-rate by defining an address transform function, which maps a VQ address of an input block which will be encoded into a new address in the reordered codebook by using side match error. In one case that an original address can be transformed into a new transformed address which is lower than the threshold value, we encode the new address of the transformed convector, and in the other case we encode the address of the original convector which is not transformed. Experimental results indicate that the proposed scheme reduces the bit-rate by 45~50% compared with the ordi-nary VQ method forimage compression, at the same quality of the reconstructed image as that of the ordinary VQ system.

• The Journal of the Acoustical Society of Korea
• /
• v.22 no.4
• /
• pp.299-307
• /
• 2003

There have been many studies on speaker normalization which aims to minimize the effects of speaker's vocal tract length on the recognition performance of the speaker independent speech recognition system. In this paper, we propose a simple vector quantizer based linear warping speaker normalization method based on the observation that the vector quantizer can be successfully used for speaker verification. For this purpose, we firstly generate an optimal codebook which will be used as the basis of the speaker normalization, and then the warping factor of the unknown speaker will be extracted by comparing the feature vectors and the codebook. Finally, the extracted warping factor is used to linearly warp the Mel scale filter bank adopted in the course of MFCC calculation. To test the performance of the proposed method, a series of recognition experiments are conducted on discrete HMM with thirteen mono-syllabic Korean number utterances. The results showed that about 29% of word error rate can be reduced, and that the proposed warping factor extraction method is useful due to its simplicity compared to other line search warping methods.

• Journal of the Korean Institute of Telematics and Electronics
• /
• v.27 no.3
• /
• pp.21-30
• /
• 1990

This paper describes a hardware implementation of the interframe monochrome video CODEC using a MC-CVQ(Motion Compensated and Classified Vector Quantization) algorithm. The specifications of this CODEC are (1) the resolution of image is $128{\times}128$ pixels, and (2) the transmission rates are about 10frames/sec at the 64Kbps channel. In order to design the CODEC under these conditions, it is implemented by a multiprocessor system composed of MC unit, CVQ nuit and decoder unit, which are controlled by microprogramming technique. And the 3~stage pipelined ALU(Arithmetic and Logic Unit) is adopted to calculate the minimum error distance in the MC unit and CVQ nuit. The realized system shows that the transmission rates are 6-15 frames/sec according to the relative motion of the video signal.

• Journal of the Korean Institute of Telematics and Electronics S
• /
• v.36S no.11
• /
• pp.89-94
• /
• 1999

In vector quantization (VQ), mean squared error (MSE) is widely used as a distance measure between vectors. But the distance between averages appears as a dominant quantity in MSE. In the case of image vectors, the coincidence of edge pattern is also important considering human visual system (HVS). Therefore, this paper presents a new distance measure using the variance of difference (VD) as a criterion for the coincidence of edge pattern. By using this in the VQ encoding, we can reduce the degradation of edge region in the reconstructed image. And applying this to the codebook design, we can obtain the final codebook that has a lot of various edge codevectors instead of redundant shade ones.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.42 no.5 s.305
• /
• pp.73-78
• /
• 2005

Images coded by DCT based compression contain several quality degradations by quantization process. Among them contrast distortion is the important one because human eyes are sensitive to contrast. In case of low bit-rate coded image, we can not get an image having good quality due to quantization error. In this paper, we suggest a new scheme to enhance image's contrast in DCT domain. Proposed method enhances only edge regions. Homogeneous regions are not considered in this method. $8{\times}8$ DCT coefficient blocks are decomposed to $4{\times}4$ sub-blocks for detail edge region discrimination. we could apply this scheme to real-time application because proposed scheme is DCT based method.