Study On The Design And Applications Of Cosine Modulated Filer Banks

With ever-increasing of the modern society's amount of information, the amount of data that digital signal processing systems need to deal with is growing. In order to reduce the amount of computation and save storage space, It is often required different sample rates and their mutual conversion in a signal processing systems. In such a case, multirate digital signal processing theory generates and develop from 70s of the 20th century. Multirate data signal processing applying technology can effectively reduce the complexity of signal processing, lower data transmission rate and reduce data storage memory.The filter banks is the important component of multirate signal processing, and it is widely used in many signal processing fields, such as speech signal processing, image processing and mobile communications, etc. So far, the design way of filter banks have a variety of methods: Design in the time domain or in frequency domain; Design at the overall of filter banks or at the single channel of filter banks.In recent years, Cosine modulated filter banks has been a great deal of concern and has been widely used, and It can be design by optimizing the low-pass prototype filter ,and gets analysis and synthesis filter banks by using fast discrete cosine transform. So the type of filter banks has the advantages of the low computational complexity and the easy design process. In this paper, the work starts around the design theory of the filter banks, and focuses on the design theory of the M-channel cosine-modulated filter banks, and introduces a design method that based on a linear iterative optimization algorithm cosine-modulated filter banks. This method requires only linear optimization to one-dimensional objective function. we can get a good performance prototype filter in accordance with this method, then get near perfect-reconstruction cosine-modulated filter banks that has smaller reconstruction error. A six-channel filter banks is designed with this method, the amplitude-frequency characteristic maps of the prototype filter and the analysis filter banks and the analysis filter banks is drawn with Matlab software.Error analysis is carried out, The result of the analysis verifies the correctness and validity of this method.Six-channel filter banks that is designed with this method is used to process an audio file and real-time audio signal on Matlab/Simulink. the output signal reconstructed the input signal better. It verifies the correctness of this method and the practicality of filter banks...