In order to solve the problem of slow processing speed of set partitio ning in hierarchical trees (SPIHT) due to its dynamic processing order that depends on the remote sensing image con tents,an improved SPIHT image algorithm for space time delay integral CCD (TDICCD) camera is proposed in this paper.Firstly,a wavelet-transformed image is decomposed into 4×4blocks and simultaneously all the bits are encoded in a bit-plane of a 4×4block.In order to realize the parallel processing mechanism,the three passes of the traditional SPIHT algorithm are re organized.The reorganized three passes are encoded in a parallel and pipelined manner.The experimental results s how that the proposed compression algorithm can work stably and reliably and has good compressive prop erty.Compared with the traditional methods,the performance of average peak signal-to-noise ratio (PSN R) for our method corresponds to that of the conventional approaches. However,it is superior to others for the throughput.The improved SPIHT algori thm can process 120million samples per second at an operating clock frequency of 100MHz,greatly improves the image algorithm speed and is very suitable for the application of i mage for space CCD camera.