论文部分内容阅读
文章结合医学图像的特点探究了一种图像压缩编码方法 :先对图像进行小波分解 ,然后针对不同层不同子图的特点对小波系数的各部分进行相应处理。小波分解后 ,低分辨率子图像的小波系数动态变化范围大 ,因而采用 BP神经网络进行自适应非线性预测编码 ,而对高分辨率子图像采用基于 Kohonen网络的自组织特征映射(SOFM)算法的矢量量化进行编码。上述压缩方法可以在保证重构图像质量良好的情况下获得较大的压缩比 ,从而可以较好的满足医学图像存储的要求
In this paper, a method of image compression coding is studied based on the characteristics of medical images. Firstly, the image is decomposed by wavelet, and then the parts of wavelet coefficients are dealt with according to the characteristics of different subgraphs. After the wavelet decomposition, the wavelet coefficients of low-resolution sub-images have a large dynamic range. Therefore, BP neural network is used to carry out adaptive nonlinear predictive coding, while the high-resolution sub-image is based on Kohonen network’s self-organizing feature mapping (SOFM) The vector quantization is coded. The above compression method can obtain a larger compression ratio under the condition that the reconstructed image quality is good, so as to better meet the requirements of medical image storage