In this paper, we propose a low bit-rate embedded video coding scheme that utilizes fast VQ with a structure of 3D set partitioning in hierarchical trees (SPIHT) algorithm to compress video data. Three-dimensional spatio-temporal orientation trees are segmented according to variety of 3D data subcube mean value in the different level wavelet decomposition, the 3D data subcube values are been scalar-quantized by means of human visual attribution at different steps, and the amount of data need coding smaller when the 3D SPIHT is applied. When this coding method applied causes much loss, a three dimension format of an Improvement Biblock Zero Tree Coding (IBBZTC). It provides comparable performance to H.263 objectively and subjectively when operated at the bit rates of 30 to 60 kbits/s with minimal system complexity. This method can efficiently remove correlation in image data, obtaining a low transmission bit stream. The fast VQ coding method proposed in this paper can achieve high compression performance, and a real-time compression using this method may be implemented.