高级检索

图像块动态划分矢量量化

A Novel Method of Vector Quantization with Automatic Image Partition

  • 摘要: 提出一种可以自动调节子图像块尺寸的矢量量化算法——图像块动态划分矢量量化该算法在进行编码前,首先对待编码子图像块及其相邻块进行相关度分析,通过预设的阈值(最大相关度)进行分类编码:如果相关度小于阈值,则对8× 8图像块进行编码;否则,对4× 4的图像块进行编码实验结果表明,相对于普通矢量量化,文中算法不但可以提高压缩率和编码速度,而且不会影响重建图像的质量.

     

    Abstract: This paper describes a novel vector quantization (VQ) algorithm called partition dynamically VQ (PDVQ),capable of adjusting the size of image block automatically. By the method, before encoding,PDVQ first analyzes the correlation of the encoding image block and its neighboring blocks, then decides the encode method according to the comparison of the correlation result with the threshold value(the maximum correlation,MC) given by experiment. If the correlation result is less than MC, an image block size of 8×8 will be encoded, otherwise, an image block size of 4×4 will be encoded. Test result shows that PDVQ can improve the compression rate and coding speed in contrast with the normal VQ, and more importantly, it will not effect the quality of the image reconstructed.

     

/

返回文章
返回