摘要
介绍了一种基于自组织特征映射(SOFM)神经网络图像压缩的传统算法.通过对传统压缩算法的分析,提出了一种新的简单易行的分类矢量量化方法.新方法采用边缘检测,主元分析(PCA),自组织特征映射来设计码书,大大提高了图像的客观指标和主观视觉效果.实验表明,此方法效果明显优于传统的SOFM图像压缩算法.
A traditional algorithm of imge compression is proposed based on the neural network of SOFM. And a new and simple sorting vector quantization method is put forward by analyzing the traditional algorithm. The new method uses edge identification, PCA and SOFM to design the codebook, which greatly improves the objective index and the subjective visual impression. The simulation shows that the new method is better than the traditional SOFM image compression algorithm.
出处
《应用科技》
CAS
2006年第6期21-23,共3页
Applied Science and Technology
关键词
边缘检测
图像压缩
矢量量化
神经网络
edge identification
image compression
vector quantization
neural network
