基于DCT域内积与方差的分形图像编码研究

定义了图像块的 DCT频域内积 ,并证明了该内积等于时域内积 ,从而可根据频域内积大小来作为分形编码中最佳匹配块的判断 ,并推导出 DCT域上恒等变换与其余 7种对称变换的关系 .将值域块分为平稳、中间和活跃三类 ,研究对于不同类型的值域块 ,求频域内积时 ,保留低频系数个数和解码图像质量之间的关系 ,并提出了基于 DCT域内积和方差的分形图像编码算法 .实验结果表明 ,本文算法能在基本保证解码图像质量的同时 ,大大提高编码速度

Study on Fractal Image Encoding Based on the Inner Product of DCT Coefficients and Local Variance

The inner product of DCT coefficients defined in this paper,which is equal to that of spatial coefficients,can be determined as close match in the fractal coding.The relationship on DCT domain wes induced between indentity transformation and the rest of the transformation.The range block wes classified into three classes:steady,middle and active.The relationship between the number of low frequency coefficients to be reserved and the quality of the decoded image is analyzed while calculating the inner product of DCT coefficients according to the different kind of range block.The algorithm of fractal image encoding based on the inner product of DCT coefficients and local variance was proposed. Experiments show that the speed can be improved greatly with the ensurance of the decoded image quality.