A Novel Filter Banks and Application in Image Compression

Based on the wavelets theory, a nonlinear optimization model is established to design a class of new filter banks which are known as quasi-orthogonal filter banks. They have mathematical properties such as finite impulse response, symmetry, linear phase, regularity and so on. The proposition that the quasi-orthogonal filter banks have only odd order regularity is proofed. Peak signal-to-noise ratio is taken as measurement of image compression quality. A number of image compression experiments are performed on four testing images with five different bit rates. The experimental results show that the proposed filter banks have a very good performance of signal decomposition and reconstruction and image compression.