How to encode a compressed image

Hey All,

After using dwt2 to transform the image, then comes thresholding, quantization and finally encoding..The dwt2 will (LL,HL, LH and HH) end up giving us values greater than one and less than zero. Thresholding is easy, but there will be two main problems in the next two stages. First, since the image is double the range is ]0,1[ so rounding the values will distort the whole matrix, even if I convert the whole matrix to unint8 I'll be losing a lot of info! Second problem, since we have a range that is >1 and <0.. this means that a combination of 256 symbols from huffman dictionary won't be enough!