This letter presents two adaptive interpolation methods based on applying an inverse gradient to conventional bilinear and bicubic interpolation. In simulations, the proposed exhibited a better performance than methods, particularly in edge regions. addition, can be used irrespective of magnification factor (MF) easily implemented due their simple structure.

In orthogonal frequency division multiplexing (OFDM) signal transmission via a nonselective fading channel, multiplicative distortion in the time domain acts as intersymbol interference (ISI) domain. The Letter proposes new compensation methods for OFDM signals, which is achieved With proposed methods, improved results can be obtained compared to previous schemes. symbol error rate (SER) of each scheme by baseband simulation.

In an interactive image segmentation, the quantity of a user‐given seed is known to affect segmentation accuracy. this Letter, we propose seed‐growing method expanding reduce bias given and improve To grow seed, supervised classification framework with geodesic distance features proposed. From single input image, support vector machine (SVM) classifier trained on superpixels image. Other non‐seed are then classified into object, background regions by classifier. experiments, proposed showed...

The order-statistics (OS) constant-false-alarm-rate (CFAR) detector has been known to have a small loss in detection performance relative the cell-averaging (CA) CFAR homogeneous situations. However, it results much less degradation nonhomogeneous situations, i.e. OS significant gain over CA environments. In general, as frequency diversity between noncoherent pulse bursts is prevalent radar systems, analysis of with integration needed. paper, for chi-square family fluctuating targets an...

In this paper, we propose a novel data augmentation technique called GenMix, which combines generative and mixture approaches to leverage the strengths of both methods. While models excel at creating new patterns, they face challenges such as mode collapse in GANs difficulties training diffusion models, especially with limited medical imaging data. On other hand, enhance class boundary regions but tend favor major scenarios imbalance. To address these limitations, GenMix integrates...

Generative diffusion models have emerged as a powerful tool for high-quality image synthesis, yet their iterative nature demands significant computational resources. This paper proposes an efficient time step sampling method based on spectral analysis of the process, aimed at optimizing denoising process. Instead traditional uniform distribution-based sampling, we introduce Beta distribution-like technique that prioritizes critical steps in early and late stages Our hypothesis is certain...

In this paper, we analyze the effects in medical images by applying various data augmentation techniques to deep convolutional neural network learning for classifying focal liver lesions abdominal CT images. We apply affine transformation-based, StyleGAN, Mixup, and Augmix-based VGG16 network, respectively, learn classify local into cysts, hemangiomas, metastases. For experiments, validate effect of through both a quantitative assessment comparing accuracy, sensitivity, specificity...

A new dedicated architecture for Viterbi scoring in hidden Markov model (HMM)-based real-time isolated word recognition systems is proposed. Because, HMMs, most states are connected to only three or fewer preceding states, the state transition matrix a banded upper triangular matrix. Using this property of HMM, we design an efficient architecture.

A postprocessor using fuzzy vector quantisation (FVQ) in HMM-based speech recognition is presented. method to reduce the amount of computation FVQ also proposed. We show that proposed results a higher rate with almost same compared conventional Viterbi scoring method.

The authors present a vector quantization (VQ) codebook design algorithm for recognition systems based on the phoneme-level hidden Markov model (HMM). In order to improve discrimination ability of HMMs, VQ codewords are extracted from HMM state segments units. An iterative method LVQ2 or MKM is proposed unified estimation codebooks and parameters. Although designed by showed larger distortions input speech spectrum than conventional codebooks, they always improved accuracy as compared...

The open-loop phase estimator (OLPE) has been used to recover the carrier for burst mode communications. However, OLPE is not optimal in presence of frequency offset. authors propose a new feedforward form restore total phase, including earner offset and/or proposed significantly less complex hardware compared with maximum likelihood (MLE) practical implementations.

In bit/digit-serial architectures the allowable sample word length is lower-bounded to synchronise feedback loops (the cycles) correctly. A systematic procedure for finding this lower bound and critical cycle that achieves are described. From information a successful schedule can be guaranteed all cases, which has been difficult using previous approaches.

The original Moose estimator suffers from large performance degradation when there is plant-filter mismatch. To alleviate this, we propose a which compensates for mismatch, and show by the Monte-Carlo simulation that proposed performs much better than in case of especially sampling interval relatively small.