基于遗传算法的OBSA多小波预滤波器

OBSA Multiwavelet Pre-filter Based on Genetic Algorithms

OBSA多小波对前置滤波器选取具有任意性。论文提出了一种基于遗传算法的OBSA多小波预滤波器构造方法。根据图像分解和重构的不同目的选择适应度函数,在适当的约束条件下,实现图像多小波变换的自适应预滤波。若图像多小波变换的目的是进行图像压缩,据此选择适应度函数,通过遗传算法方法,对不同的图像,自适应地得到前置滤波器,实验结果表明,利用此前置滤波器进行图像压缩,可以提高图像压缩比。若图像变换的目的是进行图像放大,通过文中介绍的方法,也可自适应地得到前置滤波器,放大后的图像质量较其它方法有明显的提高。

OBSA multiwavelets have randomicity as to selection of the pre-filters.This paper presents OBSA multi-wavelet pre-filter based on genetic algorithms.In this method,the fitness functions are selected for the different purposes of image analysis and synthesis,which are used to construct the pre_filters of images for OBSA multiwavelet transform (MWT)in proper constrained condition.If the purpose is to compress images,the fitness function is opted and then the pre_filter can be obtained adaptively via genetic algorithm for different images.It is shown that the image compression ratio can be improved if the pre_filter is used.And if the purpose is to magnify images through OBSA MWT,the pre_filter can be gained in the same way.Moreover,the quality of the image after magnified is superior to other methods.