• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.4
• /
• pp.1689-1703
• /
• 2020

When data is transmitted over an unreliable channel, the error of the data packet may result in serious degradation. The multiple description coding (MDC) can solve this problem and save transmission costs. In this paper, we propose a deep multiple description coding network (MDCN) to realize efficient image compression. Firstly, our network framework is based on convolutional auto-encoder (CAE), which include multiple description encoder network (MDEN) and multiple description decoder network (MDDN). Secondly, in order to obtain high-quality reconstructed images at low bit rates, the encoding network and decoding network are integrated into an end-to-end compression framework. Thirdly, the multiple description decoder network includes side decoder network and central decoder network. When the decoder receives only one of the two multiple description code streams, side decoder network is used to obtain side reconstructed image of acceptable quality. When two descriptions are received, the high quality reconstructed image is obtained. In addition, instead of quantization with additive uniform noise, and SSIM loss and distance loss combine to train multiple description encoder networks to ensure that they can share structural information. Experimental results show that the proposed framework performs better than traditional multiple description coding methods.

• Journal of Biomedical Engineering Research
• /
• v.28 no.3
• /
• pp.429-440
• /
• 2007

A coded excitation has been studied to improve the performance for ultrasound imaging in term of SNR, imaging frame rate, contrast to tissue ratio, and so forth. However, it requires a complicated arbitrary waveform transmitter for each active channel that is typically composed of a multi-bit Digital-to-Analog Converter (DAC) and a linear power amplifier (LPA). Not only does the LPA increase the cost and size of a transmitter block, but it consumes much power, increasing the system complexity further and causing a heating-up problem. This paper proposes an optimized 1.5bit fourth order sigma-delta modulation technique applicable to design an efficient arbitrary waveform generator with greatly reduced power dissipation and hardware. The proposed SDM can provide a required SQNR with a low over-sampling ratio of 4. To this end, the loop coefficients are optimized to minimize the quantization noise power in signal band while maintaining system stability. In addition, the decision level for the 1.5 bit quantizer is optimized for a given input waveform, which results in the SQNR improvement of more than 5dB. Computer simulation results show that the SQNR of a FM(frequency modulated) signal generated by using the proposed method is about 26dB, and the peak side-lobe level (PSL) of its compressed waveform on receive is -48dB.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.42 no.7 s.337
• /
• pp.21-26
• /
• 2005

An area efficient, low power switch-matrix semidigital FIR reconstruction filter for delta-sigma D/A converter is proposed. Filter coefficients are quantified to 7-bit and 7 current sources that correspond to each coefficient bit are used. The proposed semidigital FIR reconstruction filter is designed in a 0.25 um CMOS process and incorporates 1.5 mm$^{2}$ of active area and a power consumption is 3.8 mW at 2.5 V supply. The number of switching transistors is 1419 at 205 filter order. Simulation results show that the filter output has a dynamic range of 104 dB and 84 dB attenuation of out-of-band quantization noise.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.24 no.12B
• /
• pp.2382-2390
• /
• 1999

In this paper, we employed a new quantization scheme called sample-adaptive product quantizer (SAPQ) to quantize image data based on the differential pulse code modulation (DPCM) coder, which has fixed length outputs and high bit rates. In order to improve the performance of traditional DPCM coders, the scalar quantizer should be replaced by the vector quantizer (VQ). As the bit rate increases, it will be nearly impossible to implement a conventional VQ or modified VQ, such as the tree-structured VQ, even if the modified VQ can significantly reduce the encoding complexity. SAPQ has a form of the feed-forward adaptive scalar quantizer having a short adaptation period. However, since SAPQ is a structurally constrained VQ, SAPQ can achieve VQ-level performance with a low encoding complexity. Since SAPQ has a scalar quantizer structure, by using the traditional scalar value predictors, we can easily apply SAPQ to DPCM coders. For synthetic data and real images, by employing SAPQ as the quantizer part of DPCM coders, we obtained a 2~3 dB improvement over the DPCM coders, which are based on the Lloyd-Max scalar quantizers, for data rates above 4 b/point.

• Journal of Internet Computing and Services
• /
• v.8 no.3
• /
• pp.19-32
• /
• 2007

This paper present a new face recognition method based on LVQ neural net to construct a real time face recognition system. The previous researches which used PCA, LDA combined neural net usually need much time in training neural net. The supervised LVQ neural net needs much less time in training and can maximize the separability between the classes. In this paper, the proposed method transforms the input face image by PCA and LDA sequentially into low-dimension feature vectors and recognizes the face through LVQ neural net. In order to make the system robust to external light variation, light compensation is performed on the detected face by max-min normalization method as preprocessing. PCA and LDA transformations are applied to the normalized face image to produce low-level feature vectors of the image. In order to determine the initial centers of LVQ and speed up the convergency of the LVQ neural net, the K-Means clustering algorithm is adopted. Subsequently, the class representative vectors can be produced by LVQ2 training using initial center vectors. The face recognition is achieved by using the euclidean distance measure between the center vector of classes and the feature vector of input image. From the experiments, we can prove that the proposed method is more effective in the recognition ratio for the cases of still images from ORL database and sequential images rather than using conventional PCA of a hybrid method with PCA and LDA.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.52 no.4
• /
• pp.175-183
• /
• 2015

The proposed algorithm consists of two stages: the learning and synthesis stages. At the learning stage, we first collect various high-resolution (HR)-low-resolution (LR) text image pairs, and quantize the LR images, and extract HR-LR block pairs. Based on quantized LR blocks, the LR-HR block pairs are clustered into a pre-determined number of classes. For each class, an optimal 2D-FIR filter is computed, and it is stored into a dictionary with the corresponding LR block for indexing. At the synthesis stage, each quantized LR block in an input LR image is compared with every LR block in the dictionary, and the FIR filter of the best-matched LR block is selected. Finally, a HR block is synthesized with the chosen filter, and a final HR image is produced. Also, in order to cope with noisy environment, we generate multiple dictionaries according to noise level at the learning stage. So, the dictionary corresponding to the noise level of the input image is chosen, and a final HR image is produced using the selected dictionary. Experimental results show that the proposed algorithm outperforms the previous works for noisy images as well as noise-free images.