论文部分内容阅读
矢量量化（ＶｅｃｔｏｒＱｕａｎｔｉｚａｔｉｏｎ）作为一种有效的图像数据压缩技术，越来越受到人们的重视。设计矢量量化器的经典算法ＬＢＧ算法，由于运算复杂，从而限制了矢量量化的实用性。本文讨论了应用神经网络实现的基于边缘特征分类的矢量量化技术。它是根据人的视觉系统对图像的边缘的敏感性，应用模式识别技术，在对图像编码前，以边缘为特征对图像内容分类，然后再对每类进行矢量量化。除特征提取是采用离散余弦变换（ＤＣＴ）外，图像的分类和矢量量化都是由神经网络完成的。实验结果表明，和单纯用神经网络直接进行矢量量化相比，应用这种技术的图像编码压缩比和译码图像质量都有明显的提高。
Vector Quantization (Vector Quantization) as an effective image data compression technology, more and more people’s attention. The design of the vector quantization method of the classic LBG algorithm algorithm, due to the complexity of the operation, which limits the vector quantization utility. This paper discusses the application of neural network based on edge feature classification vector quantization technology. It is based on human visual system on the edge of the image sensitivity, the application of pattern recognition technology, before encoding the image, the edge of the characteristics of the image content classification, and then for each type of vector quantization. In addition to feature extraction is the use of discrete cosine transform (DCT), the image classification and vector quantization are carried out by the neural network. The experimental results show that, compared with the vector quantization directly using neural network, the compression rate and the quality of the image coding using this technique are significantly improved.