高级检索

    支持随机浏览的渐进几何压缩

    Progressive Geometry Compression for Interactive Browsing

    • 摘要: 由于三维物体的自遮挡,浏览三维网格是一个典型的随机访问问题.如果在将网格压缩后,只传输与解码当前视点下可见区域的数据,就可以节约网络带宽和解码资源.目前的网格压缩方法基本上没有考虑随机访问.提出一种基于小波变换且支持随机访问的渐进几何压缩方法,基本思想是将网格表面分成很多块,对每块的细节信息独立进行编码,然后只传输当前视点下可见块的细节信息.代表细节信息的小波系数被组织成零树,设计了一种修正的SPIHT算法来对每棵小波零树独立进行压缩.实验结果表明,该方法取得了与PGC方法相当的压缩效率,但如果只传输可见区域的细节信息,该方法需要传输的数据量只是PGC方法的60%左右.

       

      Abstract: Because of self-occlusion, browsing 3D objects is a typical random access problem. After compression, if only the data of visible regions under the current viewing parameter is transmitted and decoded, the network bandwidth and decoding resources can be saved. Almost all state-of-the-art mesh compression schemes do not take random access into account. A progressive geometry compression scheme based on wavelet transform, which supports interactive browsing, is proposed. The main idea is to partition the mesh into many blocks and encode the details of each block independently, and then transmit only the encoded details of visible blocks. The wavelet coefficients which represent mesh details are grouped into zerotrees, and each zerotree is compressed independently using a modified SPIHT algorithm. The experimental results indicate that the proposed scheme has achieved comparable compression efficiency with PGC, and the amount of data required to reconstruct only visible regions is about 60% of the complete bitsteam.

       

    /

    返回文章
    返回