A method for interference multi-spectral image coding is proposed based on region of interest (ROI) coding of embedded block coding with optimized truncation (EBCOT). After wavelet transform, the vertical wavelet decomposition is taken for the ROI which include the multi-spectral information in the high frequency subband,and the coefficients of ROI are lifted by bit-plane. The different significant weights are given to different coding passes in different bit-planes, which are encoded in the significant weights order in T1 encoder. T2 encoder gives feedback to control the coding depth according to the accumulated bit rate, and the rate-distortion optimization truncation is completed at last. The experimental results show that the proposed scheme can improve the reconstructed image quality and efficiently reduce the computational complexity and memory usage of EBCOT (for test image at bpp=1, the average peak signal-noise ratio of the whole, the ROI and the background increases more than 0.1 dB, and the average computational complexity and memory usage reduces more than 40% and 60%), which is suitable for hardware implementation of interference multi-spectral image compression system.