Данная работа посвящена вопросам исследования и разработки модели и алгоритмов параллельной обработки одномерных (звуковых) и многомерных (изображений) сигналов на основе вейвлет-функций Хаара и Добеши. Разработка параллельных алгоритмов базируется на модели многопоточной обработки данных с использованием библиотеки параллельного программирования OpenMP и TBB. Исследования по оценке ускорения процессов обработки сигналов с применением алгоритмов дискретного вейвлет-преобразования Хаара и Добеши проводились на компьютере с четырехядерным процессором.

Ушбу тадқиқот иши Хаар ва Добеши вейвлет функциялари асосида бир ўлчовли (товуш) ва икки ўлчовли (тасвир) сигналларга параллел ишлов бериш моделлари ва алгоритмларини тадқиқ қилиш ва ишлаб чиқиш масалаларига бағишланган. Параллел алгоритмларни яратиш OpenMP ва TBB параллел дастурлаш кутубхоналаридан фойдаланган ҳолда маълумотларга кўп оқимли ишлов бериш моделларига таяниб амалга оширилган. Хаар ва Добеши дискрет вейвлет ўзгартириш алгоритмларини қўллаган ҳолда сигналларга ишлов бериш жараёнларини тезлаштириш бўйича тадқиқотлар тўрт ядроли процессордан иборат компьютер негизида ўтказилди.

This paper is devoted to research and development of a model and algorithms for the parallel processing of one-dimensional (sound) and multidimensional (images) signals based on the Haar and Daubechies wavelet functions. The development of parallel algorithms is based on a multi-threaded data processing model using the OpenMP and TBB parallel programming libraries. Studies to evaluate the acceleration of signal processing using the Haar and Daubechies discrete wavelet transform algorithms were carried out on a computer with quad core processors. As the results of the study, the results of testing the performance of algorithms are given, such as the average image processing time based on the Haar and Daubechies wavelet, the average processing time of audio signals based on the Daubechies wavelet, and the acceleration coefficients of the computations of a one-dimensional signal. The corresponding evaluations of computational acceleration based on parallel algorithms using OpenMP and TBB technologies are proved. As a result of the research, a model of the algorithm for parallel processing of one-dimensional and two-dimensional signals based on wavelet functions using OpenMP was also developed. This model of the multithreaded parallel audio signal processing algorithm using the Daubechies wavelet launches the program and loads the values of the audio signal samples into a onedimensional array, creates parallel program threads for each processor core,synchronizes the threads and collects the processing results in the output array, and writes to a file for further use.