Hyperspectral Images Compression Algorithm By Multiband Calculating LAIS And Lookup Table Prediction Correction

In order to increase the compression ratio of LAIS-LUT algorithm,considering this algorithm only uses previous band to calculate LAIS and doesn't think of using other bands to get a better LAIS,a lossless hyperspectral images compression algorithm based on multiband calculating LAIS is proposed.Images from three previous bands of current band are used when calculating LAIS,and by comparing the correlations of those images in the local area of the current point,image that has the highest correlation is chosen to calculate LAIS,so a better LAIS can be achieved.Besides,in order to reduce the residuals,after the lookup table prediction value is gained,two corrections are added:the first correction is to use the weighted average of lookup table prediction value and LAIS reference value as a new prediction value;before the second correction,based on info from nearby pixels,the range of inter-band gradient value of the current point is predicted.If the inter-band gradient value calculated by the current prediction value is within this range,the second correction isn't needed;if not,the weighted average of the current prediction value and the prediction value from inter-band gradient prediction is used as the final prediction value.Adaptive arithmetic coding is used for coding.Experiments showed our algorithm has better results than the original LAIS-LUT.