支持向量机在边缘匹配矢量量化图像编码中的应用

Application of Support Vector Machines to Side-Match Vector Quantization Image Coding

针对边缘匹配矢量量化图像编码方法存在的出轨现象,提出一种引入支持向量机的方法.将分块后的样本图像输入支持向量机,建立区分边界区块与否的分类模型,并产生两种码书,对待编码区块,由支持向量机预测它的类型,选择相应的码书.实验结果表明,该方法处理灰度变化较大的边界和灰度变化平缓的非边界都有较好的效果.

To tackle the derailment phenomenon consisting in side-match vector quantization image coding technology,we introduced support vector machine.Feeding the training images into support vector machine,a module to classify edge block and non-edge block can be established,two type codebooks can be created too.For the block being coded,we estimated its type first,then chose a codebook accordingly.Experiment result showed that the proposed scheme deal two type blocks well.