Multispectral Image Compression Based On Prediction And Lifting Wavelet Transform

Multispectral technology is widely applied in many industry and military fields, such as environmental monitoring and object recognition, so multispectral images of high-precision and high-resolution are of more and more importance. Some schemes and methods of multispectral image compression and coding are reviewed in this paper, and some ameliorative algorithms are made to improve the compression result.Considering the redundant characteristics of multispectral images in space and within wavebands, an improved compression algorithm is proposed which is based on adaptive prediction with waveband sort and lifting wavelet transform. Firstly, the DPCM method is used to remove the redundancy within wavebands. To enhance the performance of prediction, the sequence of prediction is optimized by sorting the images of different waveband by correlations. At last, SPIHT coding is used to code residing image. Then, the prediction residual image is compressed by an effective lifting wavelet transform based approach which has low arithmetic operations. Simulations validate the effectiveness of the method proposed above.