HMM-Based-Correlations in Infrared Remote-Image

There are plenty of correlations in infrared remote images that intensities of edge are as a regradually changed. As a result, infrared images usually seem to be blurry. Consequently,the correlations can be utilized to perform efficiently coding. The paper analyzed the characters of typical infrared remote-images. Based on HMM, the contextual model was founded to practice the bit plan e-coding. Within this method, innerstate of wavelet coefficients could be estimated according to its probabilistic character. Experiment results showed that the proposed algorithm was effective in infrared remote-image coding while the compression ratio was 16. And the PSNR can be improved by 0.3-1.2 dB. Moreover there is rarely visual distortion in reconstructed images. Meanwhile the method would have broader application prospects.