A fast algorithm for high quality vector quantization codebook design

Abstract

In this paper we present a new theoretical approach to the problem of optimal Vector Quantization. We base, in fact, our method on the a priori explicit analysis of the effects on the MSE distortion introduced by an arbitrary exchange of training vectors among clusters. Even when the theoretical results corresponding to the simplest possible case are used, the proposed algorithm outperforms the GLA method to an impressive extent both in speed and in performance. Experiments on different images from the USC database have proved that the proposed algorithm is 5 to 10 times faster than the GLA method increasing the convergence Peak Signal to Noise Ratio (PSNR) of up to 1.12 dB.

Implementation of navigation and control algorithms in aircraft’s onboard computer systems

“…В настоящей работе, наряду с алгоритмами семейства CORDIC, рассматривается альтернативный подход реализации ал-горитмов генераций линий и поворота, связанный с использованием целочисленной арифметики. Среди множества алгоритмов, способных решить эту задачу, выделяются алгоритмы Брассини [2], которые активно используются зарубежными исследователями и инженерами в различных приложениях.…”

“…Целочисленные алгоритмы -это группа алгоритмов (алгоритмы Брезенхема, Брас-сини [2], цифрового дифференциального анализатора ЦДА и др. ), которые используют только целочисленные операции алгебраического сложения и сдвига.…”

“…В соответствии с предложением работы [2], точка с целыми коор-динатами ( , ) принадлежит прямой линии, выходящей из начала координат с коэффициентом наклона…”

A new two-step learning vector quantization algorithm for image compression

“…Braccini et al (1995) approached the problem of designing a vector quantizer from a different point of view. The authors analysed the effects on the mean square error (MSE) distortion introduced by an arbitrary exchange of training vectors among clusters and presented a training vectors reclassification strategy to solve the problem of optimal vector quantization.…”

“…However, the performance of unsupervised LVQ largely depends on how the initial weight vectors are arranged. In this paper, we introduce a decision function (Braccini et al, 1995) to determine the incorrectly clustered training vector and correct the corresponding codeword. Once the unsupervised LVQ is done, we check whether there are any incorrectly clustered training vectors in LVQ by using this decision function and correct the neural network in the second learning stage according to the correcting learning rules presented in the paper.…”